Yep, it's an actual GPS. It's a Garmin Oregon 650 designed for outdoors stuff. I did pull the GPS tracks into Google Earth last night, and there Google calculated something like 13.5 miles compared to the Garmin calculation on the GPS unit itself of 16.0. My best guess is Garmin measured every distance between every point while Google smoothed them a bit. Two and a half miles of small, extra segments seems huge, but it's the only theory I have.
My own experience with GPS units has been limited to older models with no computer connection -- or maybe there was over serial... certainly my Grandad has never had a computer to try it! That sounds like a really odd discrepancy, perhaps try importing the route into different software and see how it varies, does Garmin offer a desktop or web app that you could trail? My concern is that is actually pretty dangerous if it is reporting differently in the field, though I suppose the actual co-ordinates should be accurate... but then how would you know?