Percentage of grade - gotta love these hills!

[mudmucker]

Actually you reminded me of a point for those who don't live in the "up and down varied terrain with short distance" scenario. I would expect the data to be more accurate under less varied conditions: so if you have a long consistent, non-undulating grade I would expect that data to be more consistently accurate. And, road grade cut into the mt/hill is going to be different than the adjacent hillside slope of that part of the terrain not affected by road construction, and how accurately might that road grade be depicted based on the rest of the surrounding survey data. And if it's new road construction after the USGS survey was done.....

As for Garmin measurements, SadieKate did a number of comparisons of different cycling computers but I think this was for ascent or amount of climbing vs determining actual grade. As to her comparisons, I think she was comparing barometric devices and also measured against a barometric ciclosport device but we are back to understanding which is the correct result. She may see this and correct me if I'm wrong. Generally I recall on some other forums that people find the Garmin may register higher than what is and I think for SadieKate's area of the US she realized she might need to incorporate some data smoothing (within Sporttracks) to bring it down within a reasonable and believable number for feet climbed.