What it takes to live on Mars

Air comes first. Without it, the rest does not matter.

Mars’s air is about 95% carbon dioxide (CO₂)1. People need oxygen (O₂). The habitat has to make oxygen, scrub the CO₂ the crew breathes out, and hold pressure. The question is how much O₂ you ship from Earth and how much you make on site.

Published NASA’s Life Support Baseline Values and Assumptions Document (BVAD Rev 2) puts long-mission oxygen at 0.895 kg per person per day, in a band of 0.61–1.089. The 2022 edition is higher than the older 0.82 figure because it assumes more daily exercise.

Modeled Four people for 365 days need about 1,307 kg of oxygen (0.895 × 1,460 crew-days). That is about 149 g every hour if you make it steadily on site — roughly twelve times MOXIE’s best hour.

Measured NASA’s MOXIE experiment on Perseverance ran 16 times between 2021 and 2023, making 122 g of oxygen in total, with a best rate of 12 g/h at ≥98% purity1011. It proved the chemistry on Mars. It did not prove a year-long plant at crew scale.

Where the oxygen comes from

Modeled The thermochemistry floor for splitting CO₂ is about 4.9 kWh per kg of O₂. At 149 g/h that is about 0.73 kW continuous — heaters, compressors and losses not included. At MOXIE-like electrical efficiency (~10%), that floor becomes about 7 kW. Treat 1–7 kW as the band until a scaled unit runs.

Water is recycled, not delivered by the tonne.

Water is heavy to land. An outpost recycles almost all of it and, if needed, digs ice for makeup.

Published BVAD Rev 2 budgets 4.48 kg of water per person per day for an early planetary base (Table 4-20): 2.00 drinking, 0.50 food rehydration, and 1.98 hygiene total9.

Modeled Four people for a year need about 6,541 kg open-loop (4.48 × 1,460). With 98% recovery, makeup falls to about 131 kg a year.

Measured In 2023 NASA reported that the ISS water system reached a 98% recovery milestone13. That figure is flight hardware performance on a station resupplied from Earth. A Mars outpost must carry its own spares; spares mass is not counted in the simple yearly total.

Power runs air, water and heat. It cannot stop for weeks.

Mars gets about 43% of the sunlight Earth does at the top of the atmosphere (586 W/m² against 1,361)1. That is already a thinner budget. Global dust storms can cut sunlight further for months. Solar arrays fade in that dust; a small nuclear reactor (fission) does not, but it is heavier and harder to approve for launch.

Published A NASA Jezero trade study (Rucker et al. 2016) puts the crew-phase peak near 31 kW (text: “about 31 kW”; combined Expedition-1 loads also stated as 31,915 W ≈ 31.9 kW) for habitat, science and MAV conditioning14. An oxygen plant sized for ascent propellant takes about 34 kW in one published case (Kleinhenz & Paz 2017)15. Modeled Propellant cooling adds about 6.7 kW (Study 002 stack; not a Rucker table cell). Breathing oxygen at the high end of Study 001’s band can take up to about 7 kW. Modeled Sum: 78.8 kW. Energy per sol: 78.8 × 24.66 ≈ 1,943 kWh.

Modeled Kilopower-class units are published at 10 kWe nameplate and about 8.75 kWe delivered, at 1,500–1,800 kg each1617. Ten running units deliver about 87.5 kWe; plus one spare is 11 units and about 16.5–19.8 t. No 10 kWe flight unit has been built. KRUSTY demonstrated a 1 kWe-class fission system on the ground in 2017–201818.

Shelter holds the air in and keeps radiation out.

A habitat is a pressure vessel under soil cover. Outside air pressure is near zero compared with a cabin. The crew has never lived years at 0.38 g with a 3–22 minute delay to Earth.

Radiation · soil is the cheapest shield

Earth’s magnetic field and thick air stop most cosmic rays. Mars has almost no magnetic field and air under one percent of Earth’s pressure1. More radiation reaches the ground. Soil piled over the habitat is the usual first shield.

Measured On the ground at Gale crater, Curiosity’s RAD instrument recorded an average of 0.64 ± 0.12 millisieverts per day from galactic cosmic rays5. That is about 234 mSv in an Earth year by our arithmetic (0.64 × 365). NASA’s career limit for astronauts is 600 mSv of effective dose7. At the surface rate alone, that limit arrives in roughly 2.6 years.

Measured During Curiosity’s cruise, RAD measured 1.84 ± 0.30 mSv per day inside the spacecraft6. A round trip of two 180-day legs at that rate is about 662 mSv — already past the 600 mSv career limit, before any time on the surface. The comparison is approximate: the limit is an “effective dose,” and the RAD numbers are dose equivalent.

How much dirt overhead?

Modeled A published GEANT4/AtRIS study (Röstel et al. 2020) gives depths of rock and ice-rich ground that bring the habitat dose under 50 or 100 mSv per year8. For a 50 mSv/yr target in dry dense rock, that is about 575 g/cm² of cover — roughly 2.1 m of andesite. Scaling the same areal density to loose soil at 1.5 g/cm³ (our approximation, not a published result) gives about 3.8 m. Over a small habitat footprint of 160 m², that is roughly 920 t of cover. Ice-rich ground cuts the mass by roughly half for the same target8. Full arithmetic is in Study 003.

The shell

Published Mean surface pressure on Mars is about 0.636 kilopascals (kPa), or 6.36 mbar1 — under one percent of Earth’s sea-level pressure (~101 kPa). Cabin options in BVAD Rev 2 include 101, 70.3 and 56.5 kPa9. Outside pressure is so low that the pressure difference across the wall is essentially the cabin pressure.

Modeled For a 3 m radius cylinder at 101 kPa, the wall carries about 0.30 MN per metre of circumference (p × r). Lower cabin pressure cuts that load almost in proportion, at the cost of a richer oxygen mix and higher fire risk. Regolith cover of the kind in Study 003 (about 575 g/cm²) provides only about 21 kPa of overburden at Martian gravity — far short of balancing 101 kPa. The soil is radiation shielding, not structure. A buried habitat still needs anchors or a floor for roughly 13 MN of net uplift across a 160 m² footprint. Details in Study 004.

The body

Measured Surface gravity on Mars is 3.73 m/s², about 0.38 of Earth’s1. Whether that is enough to protect bone, muscle, eyes and the cardiovascular system over a lifetime is an open question. We do not know yet.

Measured NASA’s Twins Study compared astronaut Scott Kelly, after nearly a year in low Earth orbit, with his identical twin on the ground. Most changes returned toward baseline after landing. A few did not on the timescales studied: some gene-expression differences, DNA damage markers, eye changes, and slower cognitive-test performance that lasted six months after return19.

The mind

Measured Mars-500 sealed six male volunteers in a Moscow mock spacecraft for 520 days (2010–2011). All six finished. Individuals differed sharply: two of the six reported 85% of the perceived conflicts, and friction with “mission control” was reported far more often than friction inside the crew. As a group the crew grew more sedentary over time2021. The crew was all male; results do not extend cleanly to mixed crews. A ground run cannot reproduce weightlessness or radiation.

Other analogs — HI-SEAS habitats on Mauna Loa, NASA’s HERA campaigns, Antarctic winter-overs — point the same way: crew selection and ground support matter as much as habitat volume. See also The Psychology of Leaving Home and Life Away From Earth.

After air and water close, most of the yearly cargo is food.

Once water is recycled and oxygen is made locally, food dominates what you still ship each year.

Published BVAD Rev 2 budgets 2.39 kg of food per person per day including packaging9.

Modeled Four people for a year need about 3,489 kg. In the closed-loop case of Study 001 — 98% water recovery, all oxygen local — that food mass is almost everything still imported (~3.5 of ~3.6 t).

Launch windows open about every 26 months

Earth and Mars both orbit the Sun, at different speeds. Earth sits at 1 astronomical unit (AU) from the Sun. Mars averages 1.524 AU1, about 228 million km1. Earth goes around in 365.26 days and Mars in 686.981. Because Earth is faster, it laps Mars once every 779.94 days1. That is the synodic period, about 26 months.

The cheapest path is half an ellipse

The lowest-energy way between two circular orbits is a Hohmann transfer. Picture an ellipse that just touches Earth’s orbit on one side of the Sun and Mars’s orbit on the other. You fire the engine once to leave Earth’s orbit around the Sun, coast halfway around the ellipse, and fire again to match Mars. The same Kepler derivation appears in NASA JPL educational materials27; that is a source for the method, not a review of this page. Here it is with the fact-sheet numbers:

Modeled This idealized case uses circular orbits. Real orbits are elliptical and slightly tilted, so every window is a little different. Measured The Curiosity rover’s cruise took 254 days4, close to the textbook answer.

Talking to home takes minutes

Light travels at 299,792.458 km per second30. Across 54.6 million km1, a radio message takes about 3.0 minutes one way. Across 401.4 million km1, about 22.3 minutes. A question and its answer can take three-quarters of an hour. Mission control can’t talk a crew through an emergency in real time, so the crew has to handle it themselves.

Sols, years and seasons

A Martian solar day, called a sol, lasts 24.66 hours1, which is 24 h 39 min 35 s. A Martian year lasts 686.98 Earth days1, or about 668.6 sols by our arithmetic. Mars is tilted 25.19°, close to Earth’s 23.44°1, so it has seasons. Its orbit is far more elliptical, though, with an eccentricity of 0.0935 against Earth’s 0.01671. Mars swings from 206.65 to 249.26 million km from the Sun1. Sunlight follows the inverse-square law, so Mars gets about 45% more sunlight at its closest point than at its farthest (our arithmetic: (249.26 ÷ 206.65)² ≈ 1.45). The closest point falls near the southern summer31, the season most linked to big dust storms (see Power).

The order matters

Air depends on power. Water depends on power. Shelter sets how much pressure the air can be held at. Lose power and the other three fail in turn. That is why the power study plans for a long dust storm, not an average sol. Passive heat and insulation still matter; they do not replace the electrical load for oxygen plants, pumps and life support.

What nobody has solved yet

A skeptic should be able to read this page and say “maybe.” These open problems are why that word is the right one.

• ?
  Landing tens of tonnesRobotic landers have put about one tonne on Mars. Human cargo packages in DRA 5.0 are tens of tonnes. Entry, descent and landing at that scale is undemonstrated.
• ?
  Ascent oxygen at crew scaleModeled Making 22.7 t of liquid oxygen in 480 days is about 2 kg/h — roughly 164× MOXIE’s best hour — at about 34 kW for an O₂-only plant15. That plant has not run on Mars. Details in Study 005.
• ?
  A 10 kWe fission unit on the surfaceKRUSTY was a ground test at about 1 kWe class18. The 10 kWe building block in Study 002 is paper mass until someone builds and flies it.
• ?
  Cruise dose under a career limitMSL-like shielding already puts a round-trip cruise near or above 600 mSv67. Better shielding, faster transit, or a changed limit — none of those is settled hardware.
• ?
  Years at 0.38 gPartial gravity physiology is mostly a blank. LEO and bed-rest studies do not answer it.
• ?
  Crew health with a light-time delayAnalogs show people can finish long confinements. They do not prove a mixed crew through a medical emergency 20 minutes from a reply.
• ?
  Dust, cold and maintenanceFilters, seals, radiators and solar arrays all fight fine dust and deep cold for years without a truck from Earth.

Read the studies and check the arithmetic.

This page is a summary. The design studies hold the assumptions, arithmetic and sources.