The use of token systems has been supported across a variety of populations, but little research has evaluated the reinforcing value of token systems for individuals with autism. We used progressiveratio schedules to compare the reinforcing value of an established token system, primary reinforcement, and tokens unpaired with reinforcement. Token systems were variably reinforcing for 2 students with autism and more so than primary reinforcement for 1 student. Results support formal assessment of the effectiveness of token systems. [ABSTRACT FROM AUTHOR]