Miller and Hester's 1986 review of inpatient versus outpatient alcohol treatment studies concluded with no "justification" for inpatient treatment. Further examination of these studies revealed shortcomings such as the use of random assignment designs which excluded psychiatrically-complicated patients. Carrier Foundation's inpatient/outpatient study of private psychiatric patients with alcohol and/or cocaine dependence includes a patient-treatment matching design to address weaknesses in the existing literature. Patients with high psychiatric severity and/or a poor social support system are predicted to have a better outcome in inpatient treatment, while patients with low psychiatric severity and/or a good social support system may do well as outpatients without incurring the higher costs of inpatient treatment. Preliminary results from 183 inpatients and 120 outpatients indicated outpatients, regardless of level of psychiatric severity, were 4 times more likely to be early treatment failures (chi-square = 41.2, df = 1, p < .01). While the determination of long-term follow-up status of early treatment failures is currently underway, this finding underscores the potential risk of early treatment failure in outpatient compared to inpatient substance abuse treatment programs and the importance of addressing the issue of early attrition in conducting outcome analyses.