Experimental methods are considered defective by which Sutker, Archer, and Allain failed to replicate findings by Penk and Robinowitz showing that heroin addicts volunteering for treatment obtain significantly higher Minnesota Multiphasic Personality Inventory (MMPI) scores than addicts who do not volunteer. Research on the voluntarism hypothesis, which holds that self-reported psychopathology is greater on MMPI clinical scales among addicts who enter treatment on their own volition than among those who do not, has been plagued by differences in treatment settings, types of voluntarism, and lack of experimental designs. Experimental design deficiencies were corrected in the present study, where MMPI scores of 20 former nonvolunteering addicts who voluntarily returned for treatment were compared with 20 matched addicts who had volunteered on both first and second admissions. The voluntarism hypothesis was supported. Former nonvolunteers evidenced significantly higher MMPI scores on second admission for scales F, Mf, Pt, Sc, and Si. Volunteers did not significantly differ between first and second admissions. Such findings underscore the necessity for continuing to improve methods in addiction research. Contributions of personality in heroin addiction will remain obscured until past deficiencies are corrected. Sutker, Archer, and Allain (1979) reported a failure to replicate findings by Penk and Robinowitz (1976) showing that addicts volunteering for treatment report significantly higher psychopathology on the Minnesota Multiphasic Personality Inventory (MMPI) than addicts who have not volunteered for treatment. This failure to replicate does not fit prevailing trends in addiction research, in which several investigators have found that motivational factors,-such as voluntarism, influenced personality test responses (e.g.,