Has anyone else discovered false results when using the Compare Geometry tool to analyze the differences between two parts?

As a mold designer, it is a tool that I use often.  A customer will send us a new part model and simply tell us "I don't know what all has changed, but I need to to change the tool to make this new part".  (The typical "throw it over the wall" method of project management.)  The Compare Geometry tool is frequently used to compare the old model to the new one and show all changed areas.  PROBLEM...we are now seeing it show changes that don't exist!!!  Has anyone else come across this?