We compared rates of academic responses and problem behavior during mathematics with distributed and accumulated reinforcer arrangements for 3 students with Attention-Deficit Hyperactivity Disorder who engaged in chronic, severe problem behavior. All 3 students engaged in more academic responding and less problem behavior when reinforcers accumulated throughout the session, relative to conditions in which reinforcers were distributed throughout the session or withheld completely. We then conducted concurrent-chain analyses to evaluate student preference for the reinforcer arrangements. Two students preferred distributed reinforcers, even though this arrangement continued to produce problem behavior. One student preferred accumulated reinforcers. Our data replicate previous findings regarding the efficacy of accumulated-reinforcer arrangements, but suggest that students do not always prefer the most efficacious reinforcer arrangement.