Why don't elevation gain and loss match on an out-and-back hike?

The short answer

Elevation gain and loss are each totalled fresh from the sequence of altitude readings logged during that specific leg, not looked up from a fixed height map of the trail. Walking the same ground twice produces two different sequences of readings — even though the hillside itself has not moved an inch — so the gain measured going out and the loss measured coming back are two independent estimates of the same climb, not two copies of one number. A gap of a few percent between them is the normal result of that; it does not mean you veered off the outbound line or that the app miscounted.

What should match, in theory

On a true out-and-back — walking out to a point and returning the same way — every metre of uphill on the outbound leg is a metre of downhill on the way back over that same stretch, and every metre of downhill going out is uphill coming home. Add it all up and the outbound gain should equal the return loss, and the outbound loss should equal the return gain. That is the physical reality of retracing your own steps, and it is why the two numbers land in the same neighbourhood rather than nowhere near each other.

What it does not guarantee is an exact match, because the phrase "same neighbourhood" is doing the real work in that sentence. The two legs are recorded hours, sometimes minutes, apart, by a phone that is measuring altitude fresh at every point rather than remembering what it read the first time past.

Why the two legs don't measure alike

Altitude is read independently at every point

Nothing about a recording app remembers that a particular point was already visited on the way out. Each fix, in each direction, is its own reading of wherever the phone's GPS and any barometric sensor say the elevation is at that instant. Two passes over the same square metre of trail, taken at different times, are two separate samples of a naturally noisy measurement — they cluster around the truth, they do not have to land on it twice in a row.

Barometric drift over the course of the walk

iPhones from the 6 onward carry a barometric pressure sensor, which gives a smooth read on relative altitude change but has to be read against the current air pressure — and air pressure drifts with the weather over the course of an afternoon, independently of how high you actually are. A spot that reads at one altitude on the way out can read a few metres higher or lower on the way back, purely because the air pressure moved while you were out there, and that drift lands unevenly across the two legs rather than splitting itself neatly in half.

The noise threshold is applied to each leg on its own

Before any of it gets added up, an app ignores small up-and-down jitters below some threshold so a flat stretch does not rack up gain out of pure sensor noise. That filter runs separately over the outbound readings and the return readings, and since the two are different sequences of numbers to begin with, filtering them the same way does not force them to the same total.

The line back is rarely identical to the line out

Stepping around a muddy patch, cutting a switchback slightly differently, or simply walking at a different pace changes exactly which points get logged and where. The route is the same trail both ways, but the recorded line is a slightly different sample of it each time, and elevation gain is calculated from that sampled line rather than from the trail itself.

What it doesn't mean

A mismatch between outbound gain and return loss is not a sign you took a wrong turn, that the recording glitched, or that one leg is more trustworthy than the other. Both legs are ordinary, independent measurements of a genuinely noisy quantity, and a few percent of disagreement between them is closer to expected than a suspiciously exact match would be. If the gap is large rather than small — the return leg reporting dramatically more or less than the outbound one — the more likely explanation is a longer detour, a paused recording, or a stretch under heavy tree cover on one leg and not the other, rather than anything wrong with elevation specifically.

Things that trip people up

• Expecting the two legs to be exact mirrors and reading any gap as an error — a small gap is the ordinary outcome of two independent measurements, not a fault.
• Adding outbound gain and return gain together to estimate "total climbing for the day" without checking whether that double-counts the same hillside walked twice — it does not double the difficulty, but it is not the same as the gain on a one-way trail of equal length either.
• Assuming a bigger mismatch always means a longer detour — weather-driven barometric drift over a multi-hour hike can move the numbers on its own, with no change in where you actually walked.
• Comparing gain-versus-loss figures across two different apps or two different hikes and expecting the same tightness of match — the noise threshold and the weather on the day both vary, so how closely the two legs agree is not a fixed property of the trail.

Where TrailCam fits in

TrailCam reports elevation gain alongside distance, moving time and pace for every recorded hike, walk or ride, calculated from the route as it was actually logged rather than smoothed to force a tidy match between an outbound leg and a return one. Replaying a saved route shows the same elevation profile underneath the map, so a mismatch between the two legs of an out-and-back is something you can look at directly rather than take on faith. The app is free to download.