@avishek2411: I think you have the right idea.  But the data step after proc sort can be simplified greatly by have a SET and BY statement inside a DO UNTIL (LAST.ID) loop:     data temp; input id zip cntr $; cards; 1234 30004 USD 1234 30012 USD 1289 30056 USD 1389 45009 INR 1456 56789 INR 1456 30067 USD 1567 30099 USD 1567 67890 GBP 1567 30088 INR 1567 30092 INR ; proc sort data=temp(drop=zip) out=temp1 nodupkey; by id cntr; run; data want; do until (last.id); set temp1; by id; length cntr_list $20; cntr_list=catx(',',cntr_list,cntr); end; drop cntr; run;    The only downside is that the currency list for each id is alphabetic.  It is no longer in the order of appearance.  Have to ask @sahaji whether that is meaningful deficiency. ... View more

For loading "master" into a hash you would need "multidata" with the keys you've chosen. You then would also need to iterate through the groups with the same key as shown here: SAS(R) 9.3 Component Objects: Reference But actually for what I believe you're trying to achieve it would be much better to load the small data set into a hash. As the key variables exist in both "master" and "data1" you don't need to define the variables for the hash if "setting" master before the hash definition. Code as below should do (not tested): DATA Hash_sample (drop=_:);   set master;   if _n_=1 then     do;       declare hash first_HT(dataset:"data1");       _rc=first_ht.definekey("connection_id","credit_team_id");       _rc=first_ht.definedone();     end;   if first_ht.find()=0 then output; run; ... View more