The British scientific journal Nature reviewed science entries in Wikipedia and the Encyclopaedia Britannica (E.B). Results: few differences in accuracy.
8 serious errors were found, 4 in each encyclopaedia. Slightly more factual errors were found in Wikipedia (163 v 132 for E.B).
See article on BBC
It is pretty remarkable given than Wikipedia is entirely edited by volunteers. It also brings some faith back into the value and quality of open source knowledge, given the recent issues with Adam Curry (The original podcaster :-) and John Seigenthaler Sr.