Reject and select differ in the results

Dear authors,

I'm doing some Signal vs Background analysis .. I'm studying p p > 2 gamma 2 b as a background and p p > Z h > 2 gamma 2 b as a signal ..

I'd like to make a cut on the invariant mass of bb pair M(b[1] b[2]) .. before any cut the number of events are distributed versus
M(b[1] b[2]) from 0 to 200 GeV (peaked at 125 ) for the signal and for the background it distributed in the whole range ..

the problem that when I make select 0 < M(b[1] b[2]) < 150 , the signal events drops from 4 to 0.17 (L = 200 fb^-1) , but when I make reject M(b[1] b[2]) > 150 , the signal events do not change, it go to 3.995

I don't know which cut is correct ?

   However when I plot in the two cases M(b[1] b[2]) the plots look the same

One more thing I don't get when I make select 0 < M(b[1] b[2]) < 200, the number of events drops to 0.77 (although they are almost in this range only- so it expected to remain the same ~ 4 or 3)

ps. For any new cut I make sure any previous cuts have been removed by " remove "

Thanks,
Safinaz