Hi all, I checked the effect of the last bug fixes in recoilNtp. They are 1) fix in the size of pi0's vectors to avoid memory corruption (from ric) 2) fix in the variable tproTrk. Before it was a random number (since it was not filled at all). This variable is used in the GTVL definition The statistics used is: data 56fb-1, all generic MC available (~60fb-1) This is the comparison data(without bugs)-data(withbugs): http://www.slac.stanford.edu/~daniele/testdata/testdata-comp.html This is the comparison geneMC(without bugs)-geneMC(withbugs): http://www.slac.stanford.edu/~daniele/testgene/testgene-comp.html This is the new comparison data-mc: http://www.slac.stanford.edu/~daniele/compdatagene/compdatamc-comp.html Comments: 1) the comparison in data make sense since the differences show up in the number of charged tracks and in the Mx. This is correct. With the bug in (I mean the tproTrk one) sometimes (in a randomic way) one track was rejected and this introduced a shift down in the number od reconstructed tracks and Mx 2) the comparison in MC shows the same behaviour seen in the data but it seems to be more evident. Moreover something new is present in the lepton spectrum and in the fitted Mx spectrum (more events at high Mx). The trk spectrum is completely different now. 3) the new comparison data-mc shows a better agreement in the trk spectrum. The fitted Mx before cuts for the depleted sample show large disagreements (at high Mx) Daniele