I routinely take the breadcrumb data from the GPS application on my Pocket PC and import it into GE as a kml file. For 98% of the time it's dead on - within 10 feet - of what GE shows. The remaining 2% - well, it's usually when I'm in a ravine, under a bridge, or under very heavy tree cover. FWIW, my GPS module uses the Sirf III chipset, so my best accuracy is within 10 feet.

Keep in mind that just because your unit has WAAS capability, it doesn't mean that it's actually being used. Also, the number of satellites being received and their location has a lot to do with how accurate the GPS is. If your GPS has a stat called "DOP" you can get an idea of how accurate the position is. DOP is Dilution Of Precision. Ideally you want DOP to be under 2. That's within about 10 feet. DOPs higher than 5 are getting into the 30-foot plus errors. I owned a WAAS GPS that would experience DOPs of 10 or more under tree cover. That's near 100-feet of potential error. Its accuracy was dramatically improved by the addition of an amplified external antenna. This did result in shorter battery life, but it was worth the improvement in accuracy.

DOP gets high when the satellites "seen" by the GPS are close together. That's why accuracy is bad in narrow canyons. In fact, canyons can make things even worse if they introduce multipath reflections. This can make a GPS think you are hundreds of feet away from where you really are. I've had this happen when hiking between peaks - the GPS suddenly indicated that I was rapidly moving in large circles.

It is also possible that the satellite maps in GE are wrong for the area in question. If you really want to be certain if the problem is GE or your GPS, you can use a site like Groundspeak to find a benchmark in your area. You want a benchmark with adjusted coordinates, not scaled. Compare the coordinates of the benchmark to what your GPS reads and what GE indicates.