| Launched | December 11, 1998 — Cape Canaveral, Florida |
|---|---|
| Lost | September 23, 1999 — during Mars orbit insertion |
| Mission cost | ~$327 million (spacecraft + related program costs) |
| Cause | A mismatch between imperial and metric force units |
| Built by | Lockheed Martin (spacecraft) for NASA's Jet Propulsion Laboratory |
The Mars Climate Orbiter launched in December 1998 as part of NASA's Mars Surveyor '98 program. Its job was to study the Martian atmosphere and climate from orbit, and to serve as a communications relay for a companion lander that would arrive a few months later. After a nine-month cruise through space, it was set to fire its main engine and settle into orbit around Mars in September 1999.
That orbit insertion never happened. The spacecraft went silent exactly when it should have been establishing radio contact from the far side of Mars — and it never reappeared.
The root cause traced back to how two separate teams handled the same piece of data. Lockheed Martin, which built the spacecraft, had ground software that calculated small trajectory adjustments — the force produced by the spacecraft's thrusters — and reported those values in pound-force seconds, a US customary (imperial) unit. NASA's navigation team at the Jet Propulsion Laboratory took that data and fed it into software that expected the values in newton-seconds, the metric equivalent.
Nobody converted between the two. The navigation software simply treated the incoming numbers as if they were already in the units it expected.
This wasn't one dramatic mistake — it was a small, consistent error applied repeatedly. Over the course of the nine-month cruise, the navigation team performed a series of trajectory correction maneuvers using this mismatched data. Each correction nudged the spacecraft slightly further off its intended path.
By the time Mars Climate Orbiter reached Mars, the accumulated error meant it was on course to pass through the Martian atmosphere far closer to the surface than planned. Mission planners had designed the orbit insertion to bring the spacecraft to within roughly 150 km of the surface — close enough to use atmospheric drag to help settle into orbit, but not so close that the atmosphere itself would pose a danger.
Instead, tracking data calculated shortly before arrival showed the spacecraft was headed for an altitude of around 57 km — well below the roughly 80 km minimum altitude the orbiter was believed capable of surviving. At that depth, atmospheric pressure and heating were enough to destroy the spacecraft. It's presumed to have either burned up or broken apart during the encounter; there was no way to know for certain, because it never called home again.
In the aftermath, NASA's own mishap investigation board was blunt about the real failure: this wasn't simply "someone forgot to convert units." It was a breakdown in verification and communication. The specifications called for metric units throughout the project, but that requirement wasn't enforced or double-checked at the handoff between the contractor's ground software and NASA's navigation systems. Nobody's process caught the mismatch before it mattered — not a single reviewer, cross-check, or sanity test flagged that the numbers looked off by a factor of 4.45.
That's the detail that makes this story more than a punchline. Skilled engineers built both pieces of software, and each one worked exactly as designed — the failure was entirely in the gap between them, where nobody was responsible for confirming the units actually matched.
It's one of the most famous unit-conversion failures, but not the only one. In 1983, an Air Canada Boeing 767 — since nicknamed the "Gimli Glider" — ran out of fuel mid-flight after ground crews calculated the fuel load in kilograms while the aircraft's fuel gauge was still calibrated in pounds, during Canada's transition to metric. The math looked correct at every step; the unit was simply wrong. The pilots managed to glide the plane to a safe landing on a disused runway, but it's another case of the same underlying failure: a correct calculation built on a silent unit mismatch. For more on how imperial and metric units differ, see our Imperial vs Metric guide.
Both incidents share the same lesson: unit errors rarely look like errors. The math is usually done correctly — it's the assumption about what unit that math is in that goes wrong, and that kind of mistake is invisible unless someone specifically checks for it. It's a more extreme version of the exact problem behind smaller, everyday conversion mistakes — a recipe that goes wrong because a "cup" meant something different to the person who wrote it, or a DIY project where centimeters got read as inches.
It's also, fittingly, the whole reason a tool like this one is useful: converting between units by hand, from memory, under time pressure, is exactly the condition under which these mistakes happen. For anything that actually matters, running the numbers through the UnitDial converter takes the same amount of effort as trusting your own mental math — with a lot less risk of losing $327 million.