I didn't do any research about this but I think it's the following: If you run several t-tests to compare groups and then do a tukey-HSD you won't get the same results either. It's the same with the kruskal-wallis test. This happens because the variance that is used for computing significant differences changes when you compare all groups at once instead of only two groups. -- View this message in context: http://r.789695.n4.nabble.com/Kruskal-Wallis-test-and-kruskalmc-tp800813p4639003.html Sent from the R help mailing list archive at Nabble.com.