hi, Recently I've written another coincidence sorting program from the ASCII output of the singles.dat file because I need get some other information that GATE coincidence sorter can not provide. After I finished the program, I compared the results from my program and the results from GATE, then I noticed thare were some differences between them. Especially under the high activity condition (large amount of random coincidence). For example, with one model, using GATE output of the coincidences are:  Total  True   Random   Scatter   NECR  21761  4014    15056      2691    740 the output from my program:  Total  Ture    Random   Scatter  NECR  22600  4011     15833      2756   712 Though the differences of the NECR is only 3%~4%, but still I wanted to find out the causes. After I compared each event that was different in these two outputs and checked the original singles data file, I found out in which condition this will happen. Usually there are two kinds of situation: example 1, there are some enevts in the single events list: ID_of_enevt        Time_Stamp         8243                4.42524487908583266e-05 8249                4.42824325537103663e-05 8250                4.42823796782774862e-05 8257                4.43126886297442500e-05 in this case, if we set the timewindow as 10 ns, event 8249 and 8250 should be a coincidence, but from the GATE output, there's no this coincidence. example 2, for these events: 10593                5.66959683503792577e-05   10595                5.66958859230306874e-05   10595                5.66959047255570771e-05   10610                5.67960969563888841e-05   there are three events 10593, 10595 and 10595  in the timewindow, according the sorting criterion, if there are more 2 events within the timewindow, all these events should be rejected. However, GATE program give a coincidence for 10595 and 10595. So we can see that in both cases, the sequence of the events appears in the singles data file is not exactly according to the time that the events was detected (time_stamp). Sometimes, a single event with a larger time_stamp (later one) can be in front of another event with a smaller time_stamp (early one). It seems that when this happans, the GATE sorting program just ignores the first events. Hope this bug can be fixed in the next generation of GATE. And then GATE will provide more accurate results. Yuxuan