After reading richpips account of his everesting experience he mentioned that Garmin GPS units are notorious for over estimating mileage covered. Also on my recent borders 300 attempt my Garmin etrex venture told me I had covered 20% more distance than both bikehike and view ranger said I had. So exactly how accurate are Garmin GPS units at measuring distance?

I think it depends on how much climbing is involved. On a recent long (165km) and lumpy (3,700m) ride I noticed Strava knocked off about 2km off what the GPS said vs the final distance once uploaded.

Perhaps the Garmin allows for extra distance ridden along the hypotenuse, whereas Strava calculates a flat distance.

Maybe this explains why the HTR comes out longer than advertised?

If you leave your GPS on while stationary it'll record some distance travelled too, so you might need to factor that in.

Depends on a lot of factors - including built in barometric altimeter, quality of GPS signal, what settings you're using when post-processing, how much riding in tunnels etc.

Not all garmins are equal, there are a lot of them. I've ridden with an 800 and a Dakota on the same ride and the 800 is much more accurate from my experience.
Interestingly I've loaded duplicate rides to Strava and it interprets them differently.

I think its down to how the units sample rates are set, the navigation devices are trying to interpret changes in deviation for recording so to some degree they will straight line slight corners which all adds up over long distances.

If you require top accuracy then you can set the Garmin "sport" devices I.e. the edge series to sample every second, but the files get huge and battery life will suffer.

At the weekend my Dakota was set to record at "Normal" frequency.

I synced the barometric pressure at the start, and over the course of the ride there was only a 2 millibar difference.

One time I stopped at the top of the hill for maybe a minute. In that time stationary it recorded an extra 5 metres of altitude.

I haven't a clue how Strava does its math. It says the segment has 700 feet of climbing x 44 = 30,800 though after doing its calculations, it only gives me 29,352

The Dakota says 31,043

If distance or height is a goal, I generally factor in an extra 5% to what the Dakota says.

Edit - The mileage differnce was only 118 vs 116.5

Barometric pressure measuring for climbing is always going to throw up errors as the sensitivity of the sensor can be effected by windage too. The pinhole in some garmins causes big problems in the wet I've had mine register a descent on a climb before.
I built a pressure to sound sensor for a hang glider pilot once, with no references he couldn't tell if he was going up or down, it would work just raising it above your head.

The correction button on Strava I'm guessing should check all your data points recorded against map data and then calculate height gain, on your multiple climb ride I'm guessing the sample points were at all sorts of different random points on the way up.
Could be interesting looking at the track file in a viewer to see how varied the data points are for the same climb.

Elevation data, if Strava is using SRTM data isn't very accurate. Tiles of elevation information are broken into 90x90 metre squares, with a single elevation value for that area. It's accurate enough, but it's not very accurate.

numplumz wrote:....The correction button on Strava....

The what button...?