Twenty-three subjects, age 14 to 16 years, of whom 12 were girls, lived in a residence having an ongoing token economy. The measures of aggression were frequency of recorded punishment by staff for verbal or physical aggression, and frequency of surreptitiously recorded aggressive behaviors during daily recreation periods. Using an ABAB withdrawal design, the results showed that introduction of self-government reduced the rate of aggressive behavior observed during recreation periods but had no effect on rate of fines for aggression. Self-government appears to be effective in reducing interpersonal aggression, but the mechanism of this change is not clear. (Publisher abstract)