How long does it take to get to Mars?

Getting to Mars

Mars windows open for weeks every 26 months, the crossing takes 6 to 11 months, and landing takes 7 minutes. Why each part is hard.

TL;DR· 28 min read

Mars is reachable for a few weeks about every 26 months, the crossing takes six to eleven months for anything that means to stop there, and the landing takes about seven minutes, less time than a radio signal needs to reach Earth. Mars has enough atmosphere to burn a spacecraft and not enough to stop one, so every lander has needed a heat shield, a parachute and rockets or airbags. Counting this catalogue, 28 of 47 launches returned data from Mars and 11 of 18 attempts to reach the surface produced a working spacecraft there.

You get to Mars by leaving Earth on a curved path that arrives where Mars is going to be instead of where it is now, and that path only exists when the two planets are correctly arranged. That arrangement comes round about every 26 months, which is why spacecraft leave for Mars in clusters and then nothing goes for two years. The crossing itself takes between six and eleven months for anything that means to stop at Mars, and can be as little as four and a half for a flyby that does not. Then comes the part that has killed most of the spacecraft that failed: Mars has enough atmosphere to burn you and not enough to stop you, so every lander ever sent has needed a heat shield, a parachute and rockets, in that order, inside about seven minutes, with nobody on Earth able to intervene.

Mars's synodic period, the gap between one launch window and the next, about 26 months: 779.94 daysMars's synodic period, the gap between one launch window and the next, about 26 months
from the top of the atmosphere to the ground on a US lander; on every landing so far, shorter than the radio delay back to Earth: ~7 minfrom the top of the atmosphere to the ground on a US lander; on every landing so far, shorter than the radio delay back to Earth
attempts to reach the Martian surface that produced a working spacecraft on it: 11 of 18attempts to reach the Martian surface that produced a working spacecraft on it

A porkchop plot of Earth to Mars transfers for departure dates from the start of 2024 to the end of 2033. The horizontal axis is the departure date, the vertical axis is the length of the crossing in days, running from 100 to 850, and the colour is the launch delta-v in km/s, from 2.81 in the deepest blue to 7.81 in the darkest red. Five separate teardrops appear across the ten years, one per synodic period, each drifting a little later than the last, which is the 26-month rhythm drawn rather than asserted. A narrow blank sliver splits every teardrop: the lobe below it holds the Type I routes that sweep less than half way round the Sun, the lobe above holds the longer and usually cheaper Type II routes, and the sliver itself is the near 180 degree transfer, which is ruinously expensive because it would need a near-polar transfer plane. The plotting tool uses an approximate Lambert solver, so the contour values are illustrative; the departure energies this page quotes come from NASA Glenn's Interplanetary Mission Design Handbook instead. — A computed plot, drawn from arithmetic. The five Mars launch windows between 2024 and 2033, one about every 26 months, with the cost of leaving in colour and the length of the crossing up the side. Note that the cost rises smoothly away from the centre of each window: it is a gradient, with no sharp edge where the window shuts. Juan Luis Gonzalo, plotted with the porkchop-plot web app of the Space Dynamics Group (SDG-UPM), via Wikimedia Commons

Earth goes round the Sun in 365 days and Mars takes 687, so Earth laps Mars roughly every 26 months. The precise figure is Mars's synodic period, 779.94 days, and it is why spacecraft leave for Mars in clusters and then nothing goes for two years. The cheapest way across was worked out by Walter Hohmann, a structural engineer in Essen, in a 1925 book called Die Erreichbarkeit der Himmelskoerper. It is an ellipse that touches Earth's orbit at one end and Mars's orbit at the other: you fire once to leave, coast for months, and fire again to arrive. Because the coast is half an ellipse and takes a fixed length of time, you have to leave at the moment when Mars will have reached the far end by the time you get there. Get the timing wrong and the ellipse no longer meets anything useful. So a launch window is not really a gate that opens and shuts. It is a cost gradient. Departure energy is quoted as C3, the square of the speed a spacecraft still has after escaping Earth, and around the optimum date C3 is at a minimum and rises steeply on either side. The practical window is the few weeks in which your particular rocket can still lift your particular spacecraft. Everything else follows from that: how much mass you can send, which launcher you have to buy, whether an orbiter and a lander can travel together. In the 1973 Soviet campaign the window was expensive enough that a Proton could no longer carry both, so the USSR split them and flew four spacecraft instead of two. The Mars 4, 5, 6 and 7 page tells that story.

Not all windows cost the same, because neither orbit is a circle. Mars's eccentricity is 0.0935, about five and a half times Earth's, and its distance from the Sun swings between 206.7 and 249.3 million km, so every meeting has a different geometry. NASA Glenn's Interplanetary Mission Design Handbook plots the minimum departure energy for every opportunity from 1990 to 2045 and the values run from 7.7 to 15.5 km2/s2, a factor of two. Its table for the coming decades gives 9.144 for the 2026 window, 7.781 for 2033, which is the cheapest of the next twenty years, and 14.84 for 2037, the dearest. The pattern nearly repeats every 15 to 17 years, because that is roughly when Earth and Mars return to the same relative positions. Doubling C3 does not double the difficulty in any simple way. Converted into the burn a rocket has to make from low Earth orbit, the difference between the best and worst windows is about 3.6 against 3.9 km/s, under ten per cent, which is azmth's own arithmetic and not a published figure. But the rocket equation is exponential, so a tenth of a kilometre per second is a large slice of payload. Each window also offers two families of route: Type I trajectories sweep less than half way round the Sun and get there faster, Type II sweep more than half way, take longer, and usually cost less. Nine of the ten optimal missions the handbook tabulates for 2026 to 2045 are Type II.

The cruise is uneventful and the navigation is not. A launch vehicle upper stage is not sterilised and must not hit Mars, so the rocket deliberately aims to miss: Curiosity's Centaur left it on a path that would have passed about 40,000 km from the planet and arrived fourteen hours late. The spacecraft then removes that bias itself in a handful of corrections that get smaller as they go, from 59 minutes of thruster firing for 5.5 m/s in January 2012 down to 40 seconds in June to move the entry point by 200 km. Arriving is the expensive part, because a spacecraft reaches Mars going faster than Mars is going and must shed the difference or fly straight past, which is exactly what happened to Mars 4 in 1974 when its engine would not light. ExoMars TGO's insertion burn on 19 October 2016 ran 139 minutes and took off more than 1.5 km/s, and even that only bought a long capture ellipse. Circularising on rockets would cost about as much again, so ever since Mars Global Surveyor orbiters have used the atmosphere instead. Aerobraking dips the closest approach into the top of the air and lets drag shave a little energy off every orbit for months. Mars Odyssey made 332 drag passes over 77 days, cutting an 18.6-hour orbit to 1.8 hours and saving more than 200 kg of spacecraft mass, which is the only reason it fitted on a Delta II. TGO made 952 passes for about 1.02 km/s. Doing the whole capture in a single pass, called aerocapture, has never been flown at any planet.

Then there is landing, and this is where Mars is genuinely peculiar. Braun and Manning put it in one sentence in 2007: the atmosphere is thick enough to create substantial heating but not thick enough to produce a low terminal descent velocity. Surface density is about a hundredth of Earth's and the whole column is so thin that a vehicle is still doing hundreds of metres per second when it runs out of sky. Mariner 4 is where this stopped being an argument and became a design problem. Its radio occultation on 15 July 1965 measured surface pressure at 4.1 to 7.0 millibars against pre-flight expectations of the order of 85, and every lander anybody had drawn became too heavy overnight. The deciding number is ballistic coefficient: mass divided by drag coefficient times frontal area. Braun and Manning calculate that at Mars only entry systems below about 50 kg per square metre can reach subsonic speed on drag alone, and even then only within about 10 km of the ground. Every vehicle actually flown has been far above that, from 63 for Pathfinder to 115 for Curiosity. Without its parachute, Curiosity's entry capsule would have hit the ground at roughly Mach 1. On an airless world you need no heat shield at all and rockets do everything; in a thick atmosphere a parachute alone can finish the job. Mars demands a heat shield and a parachute and rockets, and gives you about seven minutes to use all three in the right order.

On the American missions the parachute hands the vehicle over at 55 to 90 metres per second within a kilometre of the ground, and that last kilometre has been solved three different ways. Viking's answer was throttled hydrazine engines and three crushable legs, arriving at about 2.4 m/s with 20 cm of clearance under it; Phoenix and InSight are its descendants. Pathfinder's answer, chosen because it was cheap and forgiving of rocks and slopes, was to wrap the lander in airbags and let it bounce, and the same architecture toughened for higher speeds carried Spirit and Opportunity. Braun and Manning note that airbags and similar mechanical attenuators top out near 0.6 tonnes, which is why the third answer had to exist. The sky crane holds the rockets well above the ground rather than letting them dig a crater under the payload, lowers the rover 7.6 metres on nylon cords, and lets the rover's own wheels be the landing gear; touchdown is at about 0.75 m/s. Curiosity and Perseverance both came down that way. Two landings sit outside those three: Mars 3 cut its parachute away just above the ground and hit at a reported 20.7 m/s inside a shock-absorbing shell, and Zhurong hovered on its engines while it looked for a safe spot. All of it is autonomous because it has to be. Curiosity's landing signal took 13.8 minutes to cross 248 million km and Perseverance's took 11 minutes 22 seconds, so the famous seven minutes of terror had already finished, one way or the other, before anyone on Earth knew they had begun. Perseverance added Range Trigger, which opens the parachute by position rather than speed, and Terrain-Relative Navigation, which recognises the ground during descent and can move the touchdown point by up to 600 metres.

So how often does this work? Counting from this catalogue rather than repeating the statistic that circulates: 47 launches have been aimed at Mars with a settled outcome, and 28 of them returned data from the planet or its immediate vicinity. That is 60 per cent, and it conceals a sharp change. Of the 23 launches up to Viking, 11 worked. Of the 24 from 1988 onward, 17. Of the 15 since 2001, 14, and the one failure never left Earth orbit. Landing has not improved as much. Eighteen spacecraft have entered the atmosphere meaning to reach the surface and 11 produced a working machine there, one of those only briefly, for a length of time that is still disputed, which the Mars 3 entry sets out. Nine of twelve American attempts worked, the single Chinese attempt worked, one of three Soviet attempts worked briefly, and neither of the two European attempts worked at all. Nothing has landed on Mars since Zhurong on 14 May 2021. The empty windows are part of the answer too. ESA suspended the ExoMars rover on 17 March 2022 and the window passed with nothing launched; 2024 passed the same way after ESCAPADE came off its rocket and MMX slipped. The 2026 window carries JAXA's MMX and ESCAPADE's departure burn, but not the five uncrewed Starships that had been promised for it, which SpaceX gave up in February 2026. The Starship and Mars page covers that.

Commonly wrong. Nine things are widely believed about getting to Mars that are wrong or badly incomplete. (1) That about half of all Mars missions fail. The figure is real and NASA has used it, but it is an average over sixty-five years that describes nothing happening now. Counting the launches in this catalogue: 11 of 23 worked up to Viking, 17 of 24 from 1988, and 14 of 15 since 2001, the only failure being Phobos-Grunt and Yinghuo-1, which never left Earth orbit. The correct statement is that getting to Mars used to be a coin toss and has not been one for a quarter of a century. Landing is a separate question with a worse answer, 11 of 18, and the two European attempts, Beagle 2 and Schiaparelli, both failed, though not in the same way: Schiaparelli jettisoned its parachute early and hit the ground, while Beagle 2 was found in HiRISE images in 2015 lying apparently intact with its solar panels only partly opened, having reached the surface and never transmitted a bit. (2) That the launch window is a hard gate and missing it means waiting 26 months. The 26-month rhythm is real, but the window is a cost curve rather than a door: departure energy rises steeply either side of the optimum until your rocket cannot lift your spacecraft. And a mission with a rocket to spare can leave the ground at any time and wait in space, which is what ESCAPADE did, launching on 13 November 2025 for a departure burn a year later. (3) That you launch when Mars is closest to Earth. You launch months before opposition, because you are aiming at where Mars will be after a six to eleven month coast rather than at where it is on launch day. Minimum Earth-Mars distance, 54.6 million km, is a fact about oppositions; it says nothing about transfers. (4) That the trip takes six months. Measured across every spacecraft in this catalogue that reached Mars, crossings ran from 131 days to 333. The fastest were flybys, which never had to slow down: Mariner 7 in 131 days, Mariner 6 in 156. The fastest arrival that actually stopped was Mariner 9's, 168 days. Modern missions cluster just past 200, and the long Type II routes took 254 days for Curiosity, 295 for Phoenix, 308 for MAVEN and 333 for Viking 2. Six to eleven months is the honest range for anything that means to stay. (5) That all windows are equivalent. NASA Glenn's handbook puts the minimum departure energy between 7.7 and 15.5 km2/s2 across the opportunities from 1990 to 2045, a factor of two, with a near-repeat every 15 to 17 years. The 1973 Soviet window cost 250 to 300 m/s more than 1971 and that alone forced the USSR to fly four spacecraft instead of two. (6) That aerobraking is how spacecraft slow down to enter Mars orbit. It is not. Capture is done with a large rocket burn, 139 minutes and more than 1.5 km/s for ExoMars TGO, and aerobraking happens afterwards, over months, to bring a long ellipse down to a working orbit. Doing the capture itself in the atmosphere is called aerocapture and has never been flown at any planet. (7) That the parachute is what slows a Mars lander down. It helps and it is indispensable, but it is nowhere near sufficient. Braun and Manning calculate that only entry systems below about 50 kg per square metre of ballistic coefficient can reach subsonic speed on drag alone at Mars; everything ever flown has been between 63 and 115. More than nine-tenths of Curiosity's deceleration happened on the aeroshell before the parachute opened, and without the parachute the capsule would still have hit the ground at about Mach 1. Every Mars landing has needed a heat shield, a parachute and rockets or airbags, and on the American missions the parachute is still a Viking design, qualified in 1972 and then carried forward by similarity for forty-five years, until NASA's ASPIRE sounding-rocket flights of 2017 and 2018 re-qualified the canopy Perseverance flew at Mars-relevant supersonic conditions. (8) That the seven minutes of terror are seven minutes of engineers watching in real time. Nobody watches. One-way light time was 13.8 minutes for Curiosity and 11 minutes 22 seconds for Perseverance, so by the time the entry signal arrives the spacecraft has been alive or dead on the surface for several minutes already. Every Mars landing is a recording. (9) That Mars is the hardest place in the solar system to land is a claim worth stating precisely rather than repeating. What is defensible, and is the mechanism Braun and Manning describe, is that Mars has an atmosphere dense enough to demand thermal protection and too thin to finish the job, so every Mars landing has needed a heat shield and a parachute and a powered or cushioned touchdown. Airless bodies need only rockets, and at Titan a parachute did the whole descent. But even that combination is not demonstrably unique to Mars: the Soviet Venera landers at Venus used a heat shield, parachutes and a crushable impact ring, so anyone claiming Mars is the only place that needs all three has to argue why Venus does not count, and azmth has not seen that argument made in print. Nobody has published a ranking of landing difficulty across the solar system either. Which body is hardest to land on is not a question anyone has settled, and this page does not settle it. What is genuinely unknown here is what happens next. Braun and Manning judged in 2007 that about 0.8 tonnes was the largest payload Viking-era parachute technology could put at 2 km MOLA elevation, and no spacecraft has yet landed materially more than that; Perseverance, at about 1,025 kg, is the heaviest. Note the condition, because it does the work: Perseverance landed at minus 2,570 m, four and a half kilometres below the elevation the ceiling was quoted for, so it has exceeded the mass without meeting the harder half of the test. Human missions are usually costed at 40 to 80 tonnes of surface elements, forty to eighty times the heaviest thing yet landed, which is Braun and Manning's own figure, and Venkatapathy and Hash still describe that in Astrobiology in June 2026 as a grand new challenge requiring new innovations and approaches. No American Mars entry system has flown that is not descended from Viking. Others have: the Soviet descent modules predate Viking, and Schiaparelli and Zhurong were independently developed. None of them broke the mass ceiling either, and nobody has demonstrated an entry system that could carry people.

What we know

Why the window comes back every 26 months

Earth takes 365.256 days to circle the Sun and Mars takes 686.980, so Earth laps Mars every 779.94 days, which is Mars's synodic period: about 25.6 months, or 2.14 years. NASA Glenn's mission design handbook derives the same figure independently as 779.935 days from the two orbital rates. A minimum-energy departure is only possible near the moment when Mars will have reached the far end of your transfer ellipse by the time you get there, and that phasing repeats once per synodic period.↗

Where the idea comes from

Walter Hohmann, a German structural engineer born on 18 March 1880, published Die Erreichbarkeit der Himmelskoerper (The Attainability of the Celestial Bodies) in 1925. In it he worked out the orbital manoeuvre that moves a spacecraft between two orbits for the smallest expenditure of energy, an ellipse tangent to both. It is still called a Hohmann transfer and it is still what a Mars mission flies.↗

What a window costs, and what C3 means

Departure energy is quoted as C3, the square of the speed a spacecraft still has after it has escaped Earth's gravity, in km2/s2. C3 of zero is a bare escape. NASA Glenn's Interplanetary Mission Design Handbook tabulates the optimal ballistic mission for each opportunity from 2026 to 2045: 9.144 for 2026, 8.928 for 2028, 8.237 for 2031, 7.781 for 2033, 10.19 for 2035, 14.84 for 2037, 12.17 for 2039, 9.818 for 2041, 8.969 for 2043 and 8.587 for 2045. Its Figure 1 plots the minimum for every opportunity from 1990 to 2045 and the values run from 7.7 to 15.5 km2/s2.↗

Why one window costs twice as much as another

Neither orbit is a circle. Mars's eccentricity is 0.0935, about five and a half times Earth's, so its distance from the Sun swings between 206.65 and 249.26 million km and every meeting has a different geometry. The handbook notes that Earth and Mars nearly return to their original relative heliocentric positions every 7 to 8 synodic periods, that is every 15 to 17 years, so the pattern of cheap and expensive windows nearly repeats on that cycle. The 1973 Soviet campaign is the clearest illustration in this catalogue: Lavochkin's own account says departure that year needed 250 to 300 m/s more than 1971, enough that a Proton could no longer send an orbiter and a lander together. That campaign is covered under Mars 4, 5, 6 and 7.↗

Type I and Type II

Each window offers two families of route. A Type I trajectory sweeps less than 180 degrees around the Sun: shorter trip, lower sensitivity to error, usually more expensive. A Type II sweeps more than 180 degrees: longer trip, usually cheaper. Between them sits a ridge of trajectories that pass through almost exactly 180 degrees, which are very expensive because they would need a near-polar transfer plane to reach a planet whose orbit is tilted 1.85 degrees out of Earth's. Of the ten optimal missions the handbook tabulates for 2026 to 2045, nine are Type II.↗

How long the crossing actually takes

Azmth's own computation from the launch and arrival dates of every spacecraft in this catalogue that reached Mars. The full spread is 131 to 333 days. The fast end belongs to flybys, which never have to slow down and so can accept a punishing arrival speed: Mariner 7 crossed in 131 days and Mariner 6 in 156. For anything that stops, the record is Mariner 9's 168 days, then Mars 3 at 188 and Mars 2 at 192. The modern cluster sits just past 200: Odyssey 200, Opportunity 201, Tianwen-1 202, Perseverance 203, Hope and InSight 205, Mars Express 206, Spirit 208, Mars Reconnaissance Orbiter 210, Pathfinder 212, ExoMars TGO 219. Longer Type II routes ran to 254 days for Curiosity, 295 for Phoenix, 304 for Viking 1, 308 for MAVEN, 309 for Mars Global Surveyor, 323 for Mangalyaan and 333 for Viking 2. Six to eleven months is the honest range for a mission that intends to stay. The link on this entry supports Curiosity's arrival only; every other figure here is arithmetic on launch and arrival dates held in this catalogue, and arrival is defined as closest approach for a flyby, insertion for an orbiter and entry for a lander.↗

The rocket is deliberately aimed to miss

An upper stage is not sterilised, so it must not hit Mars. Curiosity was released on a course that would have missed the planet by about 40,000 km and arrived roughly 14 hours late; the spacecraft then removed that bias itself. Its first correction on 11 January 2012 used about 59 minutes of thruster firing for a velocity change of 5.5 m/s. Its third, on 26 June, used 40 seconds to shift the entry point by about 200 km and advance entry by about 70 seconds. Six corrections were scheduled in all, three of them in the final 45 days.↗

Arriving costs more than departing

A spacecraft reaches Mars going faster than Mars is going, and must shed the difference or fly past, which is what happened to Mars 4 in 1974 when its engine would not light. ExoMars TGO's Mars orbit insertion on 19 October 2016 burned from 13:05 to 15:24 UTC, 139 minutes, and changed its velocity by more than 1.5 km/s. That buys a long capture ellipse; it does not buy a working orbit. Early missions such as Mariner 9 reached their science orbit entirely on rockets, at a total of roughly 1.6 km/s; later orbiters such as Mars Global Surveyor and Mars Reconnaissance Orbiter split the job, spending about 1 km/s on insertion and letting the atmosphere do the rest. Those last two figures are Isoletta and colleagues', so this entry links their paper; the ESA release, listed in the sources, covers only the TGO burn.↗

Aerobraking, and what it is worth

Aerobraking dips a spacecraft's closest approach into the top of the atmosphere and lets drag shave a little energy off every orbit, for months. Mars Odyssey made 332 successive drag passes over 77 days after its 24 October 2001 insertion, cutting an 18.6-hour orbit to about 1.8 hours. NASA Langley records that the propellant saved reduced the required spacecraft mass by more than 200 kg, which is what let the mission fly on a Delta II class rocket at all. ExoMars TGO aerobraked from 15 March 2017 to 20 February 2018, 952 atmospheric transits for a total velocity change of about 1.02 km/s, taking a one-sol orbit down to about two hours. ESA's flight dynamics team puts the typical saving for a low Mars orbiter at around 1 km/s, and credits the technique to Magellan at Venus in 1993 before JPL brought it to Mars with Mars Global Surveyor, Odyssey and Mars Reconnaissance Orbiter. The Odyssey figures here are NASA Langley's, from Tartabini, Munk and Powell.↗

Aerocapture has never been flown

Aerobraking takes months of shallow passes after a propulsive capture. Aerocapture would do the whole capture in a single deep pass, saving the insertion burn entirely. It has never been used at any planet. Isoletta and colleagues state it plainly in the Journal of Spacecraft and Rockets in 2021: the obstacles are the uncertainty in Martian atmospheric density and its variability, plus navigation error, because a single pass leaves no room to correct. Aerocapture was carried in the early design of Mars Odyssey and dropped for aerobraking when aeroshell packaging and mass became problems. The same paper is fairer to the technique than that summary suggests: it reports a NASA study finding aerocapture ready to be employed for science missions at Venus, Mars and Titan. Not flown is not the same as not ready. We found no report of a flight demonstration since.↗

Direct entry versus entry from orbit

Three spacecraft have entered the Martian atmosphere from orbit rather than straight off the approach. Viking 1 and Viking 2 each spent weeks circling while their sites were photographed and certified, then released the landers at an inertial entry speed of about 4.7 km/s. Tianwen-1 did the same thing in 2021: it captured into Mars orbit on 10 February, surveyed its landing region from a parking orbit for about three months, and only then released the entry capsule, on 14 May. Every US lander since Mars Pathfinder has gone straight in off the approach hyperbola without stopping, which is faster and cheaper but commits you to the site chosen before launch: Pathfinder entered at about 7.26 km/s, Spirit 5.4, Opportunity 5.5, Phoenix 5.59 and Curiosity under 6.0. The Soviet Mars 2, 3 and 6 descent modules also entered directly, separating from their carrier hours out and firing a solid motor to drop off the flyby path; Mars 7's module separated too but missed the planet and never entered the atmosphere at all. The entry speeds are Braun and Manning's; the Tianwen-1 sequence comes from azmth's own Tianwen-1 and Zhurong entries, and the Soviet descent modules are covered under Mars 3 and Mars 4, 5, 6 and 7.↗

Why Mars is the awkward case

Braun and Manning state the problem in one line: the Martian atmosphere is thick enough to create substantial heating but not thick enough to produce a low terminal descent velocity. Surface density is about one hundredth of Earth's, roughly 0.020 kg/m3, at a mean surface pressure of 6.36 millibars. What decides an entry vehicle's fate is its ballistic coefficient, mass divided by drag coefficient times frontal area. Braun and Manning calculate that at Mars only systems below about 50 kg/m2 can reach subsonic speed on drag alone, and only within about 10 km of the ground. Every vehicle actually flown has been well above that: 63 kg/m2 for Pathfinder, 64 for Viking, 65 for Phoenix, 94 for the Mars Exploration Rovers and 115 for Curiosity. Without a parachute, a Curiosity-class entry vehicle would reach the ground at roughly Mach 1. The same shortage of air puts most of Mars off limits. Braun and Manning recorded in 2007 that every successful landing had been below minus 1.4 km on the MOLA elevation scale, which ruled out most of the ancient southern highlands, where the average is about plus 2 km, and they describe fully half the surface as having been out of reach for that reason. Practice has barely moved: Perseverance landed at minus 2,570 m, InSight at minus 2,613 m and Zhurong at minus 4,099 m.↗

The parachute nobody has replaced

Every US Mars landing has used a disk-gap-band parachute descended from the Viking programme, and Braun and Manning record that the Viking-era 70 degree sphere-cone aeroshell shape has been used on every US Mars landed mission. The heat shield material is a different story and their 2007 table is now out of date on it: SLA-561V flew from Viking through Phoenix, but Curiosity and Perseverance both flew heat shields tiled with phenolic impregnated carbon ablator instead, because the entry profile and vehicle size could drive the surface to about 2,100 C. The Mars 2020 press kit puts SLA-561V on the back shell only. The shape is Viking's; the ablator on the two heaviest landers is not. Note the scope: that lineage is American. The Soviet descent modules predate Viking altogether, and ESA's Beagle 2 and Schiaparelli and China's Zhurong flew independently developed aeroshells and thermal protection. The Viking design was qualified in 1972 by high-altitude tests in Earth's atmosphere at Mars-relevant conditions, between Mach 1.4 and 2.1 and dynamic pressures of 250 to 700 pascals. Everything flown between Viking and Curiosity was then qualified by similarity to that programme. That stopped being true for Perseverance. NASA's Advanced Supersonic Parachute Inflation Research Experiments project flew three sounding rockets from Wallops Island across 2017 and 2018, deploying full-scale disk-gap-band parachutes high over Earth at Mars-relevant Mach numbers and dynamic pressures; NASA's own accounts say plainly that the ASPIRE tests of 2017 to 2018 qualified the Mars 2020 parachute. Two designs were flown, a build-to-print copy of Curiosity's canopy and a strengthened version of the same geometry. NASA's summary of the intervening decades is blunt: subscale development tests and subsonic low-altitude qualification, no new supersonic qualification, from Pathfinder through Curiosity. A follow-on campaign, ASPIRE2, was to qualify a Mars Sample Return parachute at Mach 2.1 and higher loads in early 2026; NASA's own description of it is written in the past conditional and azmth has found no record of a flight. Flown US nominal diameters run from 11.7 m for Phoenix to 21.5 m for Curiosity and Perseverance, deploying between Mach 1.1 and 2.2. Watch the convention: those are constructed diameters, and the Curiosity press kit quotes the same canopy as opening to nearly 16 m, which is the inflated width. The 21.5 m figure is NASA's, from the Mars 2020 press kits; Braun and Manning's 2007 table gives 19.7 m, the size then planned for Curiosity. Braun and Manning judged in 2007 that about 0.8 tonnes may be the largest payload Viking-era parachute technology can put at 2 km MOLA elevation.↗

The last kilometre: three answers

Every US Mars landing system so far has started from the same place, hanging under a parachute at 55 to 90 m/s within a kilometre of the ground; that range is Braun and Manning's, and their table covers the US missions only. Viking answered with throttled hydrazine engines and three crushable legs, touching down at about 2.4 m/s vertically and under 1 m/s horizontally, with 20 cm of rock clearance; Phoenix and InSight are its descendants. Pathfinder answered with airbags, cheap and forgiving, rated for 16 m/s vertical and 22 m/s horizontal and tens of bounces over rocks up to 0.5 m, later toughened to 26 m/s for Spirit and Opportunity. Braun and Manning note that airbags and similar mechanical attenuators top out near 0.6 tonnes. The sky crane is the third answer: rockets held well above the ground, the rover lowered 7.6 m on nylon cords, touching down at about 0.75 m/s on its own wheels. Those are the three American answers. Mars 3 used a fourth: its parachute was pulled aside by a small rocket just above the ground and the station hit at a reported 20.7 m/s inside a shock-absorbing shell, then righted itself on four petals. Zhurong used a fifth, a hovering powered descent with hazard avoidance onto legs, closest to the Viking answer but independently built.↗

The record, counted from this catalogue

Azmth's own count of the mission entries here, covering every launch aimed at Mars whose outcome is settled, that is every window through 2020. There have been 47 such launches; 28 returned data from Mars or its immediate vicinity and 19 did not. Of the 23 launches up to and including Viking, 11 worked. Of the 24 from 1988 onward, 17. Of the 15 since 2001, 14, and the single failure, Phobos-Grunt and Yinghuo-1, never left Earth orbit. Landing is separate and worse: 18 spacecraft have entered the Martian atmosphere intending to reach the surface and 11 produced a working machine there, one of them only briefly. How briefly is not settled: published figures for Mars 3's transmission run 14.5, 20 and 40 seconds, a 2024 reconstruction argues from the surviving print that the lander worked longer than any of them, and azmth's own Mars 3 entry leads with 14.5 seconds rather than the 20 that Braun and Manning quote. By operator: 9 of 12 American attempts, 1 of 1 Chinese, 1 of 3 Soviet and 0 of 2 European. Braun and Manning's account of the record up to 2007 agrees with ours where the two overlap: five successful US landings, Mars 2 a failure and Mars 3 twenty seconds of transmission, that last being one of several published figures rather than a settled one. They do not publish a full attempt-by-attempt tally, so this is agreement on the parts they state; it is not an independent replication of our count.↗

The windows nobody used

On 17 March 2022 the ESA Council, meeting in Paris on 16 and 17 March, unanimously acknowledged the present impossibility of carrying out the ongoing cooperation with Roscosmos on the ExoMars rover mission with a launch in 2022, and mandated the Director General to suspend it. Nothing else was ready, so the 2022 window passed empty. The 2024 window passed empty too, after ESCAPADE was taken off its rocket in September 2024 and JAXA's MMX slipped two years. Those are the first consecutive empty Mars windows since the 1980s.↗

What happened, and when

1925Walter Hohmann, an Essen structural engineer, publishes Die Erreichbarkeit der Himmelskoerper and works out the minimum-energy transfer between two orbits: an ellipse tangent to both. It is the route every Mars mission has flown since.
15 Jul 1965Mariner 4 passes behind Mars and its radio signal, bent and dimmed by the atmosphere on the way in and out, gives the first direct measurement of Martian surface pressure: 4.1 to 7.0 millibars, against pre-flight expectations of the order of 85. Kliore and colleagues report that the density, temperature and scale height are all lower than predicted. Every landing system designed since starts from this result.
1972The Viking programme qualifies its supersonic disk-gap-band parachute in high-altitude, high-speed tests in Earth's atmosphere at Mars-relevant conditions, between Mach 1.4 and 2.1 and 250 to 700 pascals. The tests are expensive enough that nothing comparable is attempted for forty-five years, and every US Mars parachute from Viking to Curiosity is qualified by similarity to this one.
20 Jul 1976Viking 1 lands in Chryse Planitia after weeks in orbit certifying the site, entering at about 4.7 km/s and finishing on throttled engines and crushable legs at about 2.4 m/s. Viking 2 follows on 3 September. They are the only landers released from Mars orbit apart from Tianwen-1's, which separated from its orbiter on 14 May 2021 after about three months of surveying the site from a parking orbit. Everything else has entered straight off the approach. The Zhurong entry covers the Chinese landing.
4 Jul 1997Mars Pathfinder enters directly from the approach hyperbola at about 7.26 km/s, the first NASA lander to skip Mars orbit entirely, and lands inside airbags built to survive 16 m/s vertical and 22 m/s horizontal impacts and tens of bounces. The same architecture, toughened to 26 m/s, later carries Spirit and Opportunity.
Oct 2001 to Jan 2002Mars Odyssey, propulsively inserted on 24 October 2001 and cleared after four checkout orbits, aerobrakes for 77 days and 332 drag passes, finishing in January 2002, taking an 18.6-hour orbit down to about 1.8 hours. NASA Langley records the propellant saved as reducing required spacecraft mass by more than 200 kg, which is what allowed a Delta II class launch. ESA's flight dynamics team lists the technique as pioneered by Magellan at Venus in 1993 and then used at Mars by Mars Global Surveyor, Odyssey and Mars Reconnaissance Orbiter; ExoMars TGO makes it a full year from 2017.
6 Aug 2012Curiosity flies the first guided hypersonic entry at Mars, steering its lift vector with an autonomous algorithm to shrink the landing ellipse to about 20 km, then lands on the sky crane. Entry at about 5,900 m/s, peak heating 75 seconds in at about 2,100 C on the heat shield, parachute at 254 seconds and 11 km, touchdown at about 0.75 m/s at entry plus 416 seconds. More than nine-tenths of the deceleration happens on the aeroshell before the parachute opens.
19 Oct 2016Two arrivals on one day show both halves of the problem. ExoMars TGO's insertion burn runs 139 minutes and takes off more than 1.5 km/s, putting it safely into a capture ellipse. Its passenger Schiaparelli, entering directly, jettisons its parachute early and is destroyed on impact.
Oct 2017 to Sep 2018NASA's ASPIRE project flies three sounding rockets from Wallops Island, in October 2017, March 2018 and once more in the second half of 2018, and deploys full-scale disk-gap-band parachutes high over Earth at Mach numbers and dynamic pressures relevant to Mars. NASA's own documents give the third flight as July and as September 2018; azmth has not resolved which. Two canopies are tested, a build-to-print copy of Curiosity's and a strengthened version of the same geometry, and NASA records that the campaign qualified the parachute Perseverance would fly. It is the first supersonic Mars parachute qualification programme since Viking's, and it is why the claim that nothing has been re-qualified since 1972 no longer holds for the US.
18 Feb 2021Perseverance lands with two techniques never flown before: Range Trigger, which fires the parachute mortar based on where the vehicle is rather than how fast it is going, and Terrain-Relative Navigation, which photographs the ground during descent, matches it to an onboard map and can move the touchdown point by up to 600 m. One-way light time that day was 11 minutes 22 seconds.
17 Mar 2022The ESA Council suspends cooperation with Roscosmos on the ExoMars rover after Russia's invasion of Ukraine. Nothing else is ready and the 2022 window passes with no launch to Mars. The 2024 window passes the same way, the first consecutive pair of empty windows since the 1980s.
13 Nov 2025ESCAPADE lifts off on a Blue Origin New Glenn from Launch Complex 36 at Cape Canaveral at 3:55 p.m. EST, a full year before the window it is aiming at. Azmth's own ESCAPADE entry, which reads its trajectory from JPL Horizons, records a wide loiter loop that keeps the pair in Earth's neighbourhood, passing through the Earth-Sun L2 region around late February 2026 but nowhere near it the rest of the time (on 23 August 2026 they sat about 88 degrees off the L2 direction), a trans-Mars injection burn planned for November 2026 and arrival in September 2027. It is the clearest demonstration that the 26-month rhythm constrains departure from Earth's neighbourhood rather than departure from the ground.

In pictures

One spacecraft photographing another while it falls. On 5 August 2012 Pacific time the HiRISE camera on Mars Reconnaissance Orbiter caught Curiosity descending toward the etched plains just north of the dunes that fringe Mount Sharp, while the same orbiter was listening to the lander's transmissions. The wide frame is the floor of Gale crater; the white box marks the spacecraft, and the inset is the same patch stretched to stop the bright hardware saturating. In the inset the parachute canopy and the backshell hanging below it are separately resolved, which is the entire architecture of the page in one exposure: a heat shield already gone, a parachute still doing its work, and rockets yet to fire. — A real photograph, with nothing added by an artist. Curiosity under its parachute over Gale crater on 5 August 2012, caught from orbit by HiRISE while that same orbiter was relaying its telemetry. The inset separates canopy from backshell. Credit: NASA/JPL-Caltech/University of Arizona, HiRISE on Mars Reconnaissance Orbiter; catalogued by JPL as PIA15980.

A still pulled from the video shot by a camera on the descent stage as Perseverance was lowered into Jezero crater on 18 February 2021. The rover hangs beneath on the sky crane's cords with its wheels already deployed, and the ground below is visibly disturbed by the descent engines firing above it. This is the third of the three American answers to the last kilometre: instead of landing the rockets and then getting a rover off them, the descent stage holds station overhead, pays the rover out on nylon cords and lets the rover's own wheels be the landing gear, touching down at about 0.75 m/s. Nobody on Earth watched it happen. One-way light time that day was 11 minutes 22 seconds, so the landing had already succeeded or failed before the entry signal even arrived. — Perseverance on the sky crane's cords over Jezero crater, 18 February 2021, photographed by the descent stage hanging above it. A real camera frame, and nobody saw it live: the radio delay that day was 11 minutes 22 seconds. Credit: NASA/JPL-Caltech (PIA24428).

Curiosity's heat shield, 4.5 metres across, with a worker standing at its edge for scale. NASA calls it the largest ever built for entering any planet's atmosphere, and more than nine tenths of the braking happened on it rather than on the parachute. Credit: NASA/JPL-Caltech/Lockheed Martin (PIA12117).

Onboard camera frames, published by NASA as a short animated loop, of a full-scale disk-gap-band parachute inflating high above Earth on the third and final flight of NASA's Advanced Supersonic Parachute Inflation Research Experiment. NASA's page for this clip dates the flight to 7 September 2018; other NASA documents give July 2018 for the same flight, and this page does not resolve which is right. A Black Brant IX sounding rocket launched from Wallops Flight Facility in Virginia carried the payload high enough that the canopy could be deployed at Mach numbers and dynamic pressures resembling those at Mars. NASA records this as the fastest inflation of a parachute this size on record, four tenths of a second, at a peak load of almost 70,000 pounds of force. ASPIRE matters to this page because it is the first supersonic qualification of a Mars parachute since the Viking programme's tests of 1972: everything flown in between, from Pathfinder through Curiosity, was qualified by similarity to Viking rather than by flying. — Real camera frames, looped, and nothing in them simulated: the Mars 2020 parachute inflating on a sounding rocket flight, which NASA's page for this clip dates to 7 September 2018. It was the first supersonic qualification of a Mars parachute since Viking's in 1972. Credit: NASA/JPL-Caltech.

A full-size test model of the airbag system that landed Spirit and Opportunity, inflated inside a test facility, with two people beside it for scale, one of them standing at the foot of a step ladder that does not reach the top of the stack. Each rover carried four bags of six lobes each, sewn from layers of Vectran, held to the lander not by the fabric but by a web of cords that gives the lobes their shape. The Pathfinder version was rated for 16 m/s vertical and 22 m/s horizontal impact and tens of bounces over rocks up to half a metre; the rover version was toughened to 26 m/s. JPL's own account of the campaign is blunt about why it existed, in Tom Rivellini's words, that the point was to break the bag on Earth rather than on Mars. Roughly sixty drop tests were run from 2001 to 2003 in the vacuum chamber at Plum Brook Station, where a 535 kg test article was fired by bungee cord onto a rock-studded platform at landing speed. — A full-size test airbag for Spirit and Opportunity, with two people and a step ladder beside it for scale. Cheap, forgiving of rocks, and good only to about 0.6 tonnes of payload, which is why a third answer had to be invented. Credit: NASA/JPL-Caltech, originally published by JPL as a Mars Exploration Rover mission image.

Tap a photo to enlarge.

Sources

Checked on 24 August 2026. Where the science is unsettled this page says so rather than picking a winner.

All explainers →All Mars missions →Mars numbers and data →azmth Mars: the planet →

01What we know

02What happened, and when

03In pictures

04Sources

What we know

What happened, and when

In pictures

Sources