I found that the pi0 block was defined in the wrong way, with dimension 100 instead of 150. This also applied for all the studies done for the conferences. There are a few events in 10**3 where there are actually more than 100 pi0s and therefore the memory gets screwed up. I have no clue how big the effect is, but please use VubAnalysis rf-082302 for future studies ciao ric