Sigurd's solution gives you what you were looking for but, as usual, non-sql solutions are also available. I've listed one, below, but first want to point out a problem of which you should be aware, namely that you are attempting to merge records where duplicates exist in both files.

As a result, you are probably not getting the desired resulting file.

Art

data tmp last; retain count; merge data_A (in=tv) data_B (in=td); by ID DT; if first.dt then count=1; else count+1; pattern = tv + 10*td; output tmp; if last.dt then output last; run; proc print data=last; run; ---------- "J. Man" <jman_200045@hotmail.com> wrote in message news:db3fd05.0401021249.1a4d9e13@posting.google.com... > How do I get the frequency of duplicate records for BY variables. In > the following example, I prgram to give "data_A has THREE observations > for 102-06Aug01 and data_B has TWO records for the same BY ID DT." > similarly for other records. > > Am I mising an option ? Any pointer appreciated and greetings for the > New Year to all. > > J. Man > > **** program starts; > > data data_A; > input ID DT $ VNO $; > cards; > 101 25May01 v1 > 102 30Jul01 v2 > 102 30Jul01 v3 > 102 06Aug01 v4 > 102 06Aug01 v5 > 102 06Aug01 v6 > ; > run; > > data data_B; > input ID DT $ DNO $; > cards; > 102 30Jul01 d1 > 102 06Aug01 d2 > 102 06Aug01 d3 > 102 13Aug01 d4 > 103 24Aug01 d5 > ; > run; > > proc sort data=data_A; > by ID DT; > run; > proc sort data=data_B; > by ID DT; > run; > > data tmp; > merge data_A (in=tv) data_B (in=td); > by ID DT; > pattern = tv + 10*td; > run; > > proc print data=tmp; > run;