Hi, the mX resolution plots I made on the ICHEP signal MC (so what I call mix is the mix as we got it from the generator, with the nonres component for mX>1.264GeV) can be found here: http://www.slac.stanford.edu/~kerstin/vubfit/VR011105.pdf Each plot is for 310 MeV, the black plots are for mxhadfit and the orange dashed for mxhad. I applied the following cuts: pcms>1.&&TMath::Abs(mm2)<1.&&mes>5.27&&(brecoqual==1||brecoqual==5||brecoqual==7 and I used mxhadgenwoph as the true mX. The bottom left plot is the sum of the three other lower plots. I checked for the loss of signal when using brecoqual for subtracting the combinatorial background. What we do for the evaluation is: we take the mX spectrum from signal MC after all cuts and with the comb bkgd subtracted by fits to the mES distributions and the same spectrum, but the comb bkgd subtracted using the brecoqual variable and we take the ratio of these two spectra. We see that the ratio is consistent with a constant and we fit 1.0475+-0.0410 (a little bit dependent on mb and a, but the picture is the same, the ratio is mX(mES)/mX(brecoqual)). I remembered a larger effect, but this was in the older ntuples. There are plots in BAD 894 p. 14ff that show the ratio, also split up by multiplicity categories. Cheers, Kerstin