We compare in a laboratory experiment two audit-based tax compliance mechanisms that collect fines from those found non-compliant. The mechanisms differ in the way fines are redistributed to individuals who were either not audited or audited and found to be compliant. The first, as is the case in most extant tax systems, does not discriminate between the un-audited and those found compliant. The second targets the redistribution in favor of those found compliant. We find that targeting increases compliance when paying taxes generates a social return. We do not find any increase in compliance in a control treatment where individuals audited and found compliant receive symbolic rewards. We conclude that existing tax mechanisms have room for improvement by rewarding those audited and found compliant.