The methodological difficulties in comparing the effectiveness of alternative alcoholism treatment programs were addressed. An interrupted time-series design and MMPI scores were employed with 408 men veterans to test the effectiveness of four changes in an 8-week inpatient alcoholism treatment program: The program was given its own ward; the amount of time patients spent in common activities was increased; the staff: patient ratio was improved; and contract with the outside community was broadened. The scores of MMPI tests administered at treatment termination were available for the periods of 3 yr prior to and 3 yr after the change. The results of the time-series analysis indicated that the new program was indeed more effective than the old one because nine of the 13 subscales used and the Disturbance Index (a weighted combination of all the subscales) showed a reliable decline coincident with the introduction of the new program. However, the results also showed most of these effects to be short-lived. Implications of these results and issues involved in implementing this design are discussed.