Hi Oliver, I have check your observation on misid. I have added the good track loose selection to find back Urs's results. I show plots on 2000 generic B0. The plot done on all tracks is http://www.slac.stanford.edu/~rfaccini/phys/ISL/allgt.eps I would say that I find back urs's result the correponding plot for only tracks on the recoil is http://www.slac.stanford.edu/~rfaccini/phys/ISL/recoilgt.eps This misid is twice as big and this makes sense because we are making the Breco candidate vetoing on electrons and muons. So the Breco tracks are very unlikeli to be called electrons (some can because we are using a different selector). So I do not see anything wird in this. The only oddity is why we have a misid of ~1% on the recoil tracks, but the scaling with "all tracks is clear ) Anyhow I find that 19% of the tracks above 1 GeV belong to the recoil. Do you agree on the number?This fraction goes up to 50% if one removes the cuts on momentum because the track from reco are stiffer by contruction. hope this helps ciao ric P.S. the code I used is in ~rfaccini/myRoot/test.C at slac