Twenty mildly and moderately retarded boys (age 12 to 18 years) participated in a behavior modification program in which adaptive behavior was rewarded through tokens or privileges, while maladaptive behavior was controlled through the use of fines and removal from the situation. The preliminary results (based on four months of operation) showed the token economy methods to be effective with this population, although a rapid loss of acquired good behavior during holidays raised questions regarding the durability of such results. A description of the methods, subjects' characteristics, and behavior patterns is included with the preliminary report. (RD)