Hoi Alessio, I think the chi2/ndof calculation for the tree-based comparisons is confusing (at best), especially when comparing with the numbers from the histogram based comparisons. Looking e.g. at the lower plot of http://www.slac.stanford.edu/BFROOT/www/Physics/Analysis/AWG/InclusiveSL/islrecoil/datagene_ChB0909_80fb-1/comparison7nchgallcuts.ps one would conclude from the number printed on the plot that the chi2/ndof is quite ok. Looking at the plot it is obvious that with only one point showing agreement within one sigma and all the other FAR off, the number cannot be correct. I would suggest that this improvement should be the next task on your to-do list (at least when working on this topic). Cheers, --U.