Hi all, the bad behaviour of the generic MC I mentioned in my previous mail was due a wrong removal of the candidates that belong to the reconstructed B. A large fraction of the generic MC in the run range 750000-1000000 is affected by this problem. The format for the tag word that handles the overlap seems to be the old one for a fraction of the rootfiles. I don't understand how this can happen. The only explanation is that we are running different versions of code and this is really dangerous. Alessio, may you check what is going on? Daniele