Hello,   I would proceed this way : 1/ Standardize names in both datasets (case, suppress special characters, ...) 2/ merge rows of the datasets if one standardized name contains the other 3/ Flag unmatched rows and matches for which names differ and examine flagged rows to see if some additional criterion could improve the results 4/ Repeat from 1 with an enriched standardization marco until the remaining flagged rows can be handled manually.   Of course, if the differences between names in both datasets are important this can be a huge task.   %macro stdz(name); prxchange("s/[^\w]//",-1, tranwrd(upcase(&name.),"SAINT","ST")) %mend; data main; id=_N_; set sample_main; run; proc sql; CREATE TABLE matches AS SELECT main.id, match.* FROM samples_matches match LEFT JOIN main ON (find(%stdz(main.facility_name), %stdz(match.facility_name)) OR find(%stdz(match.facility_name), %stdz(main.facility_name))) AND main.facility_state=match.facility_state AND main.facility_zip=match.facility_zip ORDER BY main.id ; quit; data want; merge main (in=inmain) matches (rename=(facility_name=match_name) in=inmatch); by id; FLAG_MAIN_ONLY=not inmatch; FLAG_MATCH_ONLY=not inmain; FLAG_FUZZY_MATCH=inmain and inmatch and facility_name ne match_name; run; ... View more