Or make a format from Data2 and use that to add the label. Data makeformat (keep=start label fmtname);      set data2;      rename (variable name for area code not mentioned)=start (variable name of area title)=label;      fmtname = "$AreaLabel"; /* or what every you like*/ run; proc format cntlin=makeformat library=(library you would like to keep it in); /* make sure the library is in the format search path. If you use a permanent library then the code above only needs to be run when you have a change in your area titles. Since those codes look like counties then they aren't likely to change often*/ use as : data want;    set data1 (or what ever your first data set is named);    AreaTitle = put(substr(id, 4,8),$AreaLabel.); run; BTW, I would be careful about calling FIPS codes area code, you will generate confusion. ... View more

You've never (that I remember at least) mentioned how big the datasets are.  How many records would you guess would be in all 50 files combined? ... View more

Possibly more effecient in some situations: Proc sql;      Create table nodup as      select distinct *      from yourdataset; quit; ... View more

How About: Data set1; infile datalines delimiter=','; input city1 $ city2 $ pass; datalines;          chi ,   det ,   3 chi ,   ks  ,   4 la  ,   dc  ,   5 la   ,  chi ,   6 ; run; Data set2; infile datalines delimiter=','; input city1 $ city2 $ pass; datalines;          chi ,   det ,   13 chi ,   ks  ,   14 la  ,   dc  ,   15 la   ,  chi ,   16 la   ,  ho ,   20 ; run; proc sql; create table want as   select * from set1    where cats(city1,city2) in (select cats(city1,city2) from set2)   union all    select * from set2    where cats(city1,city2) in (select cats(city1,city2) from set1) ; quit; Ksharp ... View more