Twelve behaviors selected for reinforcement among 16 chronic psychiatric inpatients were divided into four classes: (a) personal hygiene, (b) personal management, (c) ward work, and (d) social skills. A token economy program was introduced for each class in a sequential, cumulative, multiple-baseline format. Corrections were included for methodological deficiencies frequently enountered in past studies. Treatment variables were systematically monitored, and target behavior rates, levels of global individual functioning, general ward behavior, and off-ward behavior were assessed during baseline, implementation, and probe periods. Results indicated abrupt and substantial increases in performance of most target behaviors, significant improvements in global individual functioning (p less than .025), positive changes in general ward behavior, and increases in social interaction during off-ward activities. The findings provide strong evidence for the efficacy of a token economy and indicate that the multiple-baseline design can be a useful method for evaluating token economy programs.