"Add surface information" - "Sampling distance"?

[ValentinMüller]

I have some problems with the "Add surface information" tool. I have a drainage network as polyline and a discharge raster layer. When trying to add the mean discharge along the line as attribute to the polyline with that "add surface information" tool the results seem to be biased. It appears as cells with smaller values are getting more weight for averaging then cells with large values. In addition the surface length appears to be more or less random. The coordinate system of both layers is UTM.
I already tried to adjust the sampling distance parameter, however this didn't change much. Also I am not sure which unit this parameter has. Is it meters? Should be with UTM, right? Or is it maybe a multiplier to the cellsize? Unfortunately the tool "help" isn't much of value.
The longer I am working with version 10 of Arcgis the more I am getting the impression ESRI shipped an early beta as "final" version...

[ValentinMüller]

Hmmm. Am I right that the "add surface information" calculates the length from the raster layer itself, assuming it is always an elevation raster with the same units than the projection? This would mean I can't use the "add surface information" for any other data, right? Does anybody know any alternatives for calculating the the mean cell value along a polyline and adding the result as attribute?