Went with the ff. code: Let me know what you think and a more efficient way is much appreciated. 🙂   data want;    set have;    max_dte = max(of vardate1-vardate5); run;   proc sort data=want out=want2;    by group1 group2 group3 descending max_dte descending varid; run;   data want3;    set want2;    by group1 group2 group3 descending max_dte descending varid;    if first.max_dte; run;   proc sort nodupkey data=want3                                  out=want4 (index=(indx = (group 1 group2 group3)/unique)); by group1 group2 group3; quit;     ... View more

Thanks a lot for all your help. I still can't make it work though. Modifying a bit so that it sort of represents the data wherein Group is actually 3 variables. And yes, those are SAS numeric dates. Also looking for the most efficient way as this will run through millions of records.   Example Data: Group1, Group 2, Group3, VarID,VarDate1,VarDate2,VarDate3,VarDate4,VarDate5 A,AA,AAA,15,Oct2018,Oct2018,.,.,Oct2018 A,AA,BBB,2,.,.,.,.,Jan2019 A,AA,CCC,3,.,.,Sept2018,.,., A,AA,DDD,14,.,.,.,.,Dec2018 B,AA,AAA,45,July2018,.,.,.,July2018 B,AA,BBB,46,.,.,.,.,.July2018   Result Data should be: Group1, Group 2, Group3, VarID,VarDate1,VarDate2,VarDate3,VarDate4,VarDate5 A,AA,BBB,2,.,.,.,.,Jan2019 B,AA,BBB,46,.,.,.,.,.July2018 ... View more