Yes, this is likely to happen if your cases population is not distributed the same way as your control population. Groups that are abundant in the cases but rare in the controls will generate multiple matches. Notice that duplicate matches occur across age groups, not within the same age group. I don't know how to prevent this completely. One way of reducing the likelyhood of duplicates would be to favour exact age matches over approximate ones. To do this, you could try changing variable dist expression to: (U.age ne C.age) + (abs(U.age-C.age) > &bestAgeDiff.) + (abs(U.age-C.age) > &maxAgeDiff.) as dist, Best of luck. PG ... View more