I recently downloaded Strava, and noticed, that the mileage is different from my Bontrager speedometer/odometer. I double checked the settings on the Bontrager(wheel size)it was correct, Any idea which I should assume is more accurate?

Neither is.

You expressing OCD on Precision? You get one of those special wheel things then walk it, to lay out your race course..

things to get your 10K to be just that..

Your wheel magnet speedometers must be calibrated. Usually lay out a tape measure that is long enough for a couple of wheel revolutions. Weight the bike, and roll out the distance traveled in a couple of turns.

I still have difficulty getting the accuracy greater than about 1/2 cm. And, as the tire wears, it should get smaller, but I've also had difficulties verifying as it falls close to the limits of my measurements.

What about those micro swerves one makes with pedaling, and to keep the bike upright?

Still, I'd trust the magnet based sensors over Strava. I have one ride were I accidentally left Strava on overnight, and got about 300 miles and 14,000 feet climbing for doing nothing.

I do think Strava and RWGPS are close on distance. Give or take a couple of percent. Climbing is much more difficult to calculate, and one sees widely varying estimates.

The most accurate distance measurement is a good GPS unit. A properly calibrated cyclometer is a close second.

Since you can't calibrate Strava, calibrate your cyclecomputr to match.

When I lived in Raleigh, NC there was a measured mile near by. I assume it was used to calibrate taxi, and other official odometers. I rode it every so often to compare my bike computer mileage. The closest I got (using a cat Eye computer with only 3 digits of setting) was a couple of handfuls of wheel rotations plus or minus. I would reset the computer a digit or so and reride, only to have the error go in the other direction. I ended up deciding that this was both close enough and really a moot point. The body is more concerned about how much time and how frequently one rides and not so much how far. Andy

Given the choice of precision or accuracy, I'd rather have precision.

Phone-recorded Strava's WAY off all the time. We get some amusing results both phone recording Strava on the tandem and comparing segment times and speeds.

I was recently setting up a bike computer. I had a hell of a time getting the roll out of the front wheel... because the Siamese kept playing with the end of the tape measure

I've got to think that the wheel-measured distance includes a "wobble factor" in that you are not on a perfectly straight line (in fact, it's impossible to ride a bike in a perfectly straight line - its not stable). So the "measured mile" would be a mile plus a fraction of a percent to cover the wobble. A GPS-measured distance will not have the resolution to account for the wobble. Civilian GPS has a horizontal confidence interval of 7.8 meters and worse than that vertically. To get better than that, I think you need a GPS unit that is Defense-Department accurate. But even the US newest and most accurate nuclear weapon (the B61-12) has a Circular Error Probability of 30 meters*.

*I guess that this adds another category, besides horseshoes and handgrenades, where "close" is good enough.

Your speedometer/odometer measures the length of your actual trajectory. If you deliberately ride in a zig-zag trajectory your speedometer/odometer will measure the length of the zig-zag.

Strava uses GPS data. It knows that GPS location data is constantly affected by random errors (noise). Strava knows that because of that noise your GPS trajectory will always zig-zag a bit, even if you were riding in a perfectly straight line. For this reason, it always "smooths" the received GPS trajectory. When it sees any relatively small[-ish] zig-zags in your trajectory, it cannot tell whether this is GPS noise or you actually deliberately riding in zig-zag fashion. So, it always assumes it is just noise. It always unconditionally straightens/smooths your trajectory.

As a result, when you ride in more-or-less erratic, zig-zagish pattern, Strava results will typically measure shorter distance than your onboard Bontrager speedometer/odometer.

Well, if your Bontrager wheel circumference data is indeed accurate, than physically it is your Bontrager that's more accurate.

However, cycling is by nature a "trajectory-independent" sport: in order to "win" you have to get from point A ("start") to point B ("finish") as fast as possible. The distance between A and B is known in advance and all that matters is your timing. Nobody cares what trajectory you take between A and B, as long as you don't cut corners illegally. You don't get extra credit for riding zig-zags (and thus riding a longer distance). In that regard Strava might be more "accurate" in a sense that it is better aligned with the nature of the sport.

Exactly, and IME, Strava is pretty good at that. I also want my real-time computer to match my century route map so I don't miss turns - I don't really care if I rode 32.7km or 33.1km, and I don't care about the difference between riding a straight line and a wobbly line.

I suspect that it depends.

Any wheel driven odometer depends on how accurately you have calibrated the instrument and how precisely you ride the exact route. Any tiny error in calibration gets added over and over again so the longer the course, the more your distance will be off. In my road running days I was surprised by how much the elite runners cut every corner. On a curvy course, that can add up to several percentage points of difference.

A GPS determines distance by triangulating satellites in space. Over a very short distance I'm thinking it will be less accurate than a wheel driven odometer but, as the distance increases, the error between the starting and ending points only gets counted once.

Here's my question: "How accurate do you need it to be?"