Early attrition is a significant problem in the treatment of cocaine dependence, but it is unclear why some patients succeed in treatment while others relapse or drop out of treatment without a demonstrated relapse. The goal of this study was to determine whether baseline levels of select hormones, including the adrenal hormone and excitatory neurosteroid dehydroepiandrosterone sulfate (DHEAS), would distinguish between treatment outcome groups. Based on the literature, completion of 90 days of treatment was established as a key outcome variable. Quantitative urine levels of the cocaine metabolite benzoylecgonine (BE) and other substance of abuse analytes, plasma levels of DHEAS, DHEA, cortisol, and prolactin, and the profile of mood states (POMS) were serially measured in 38 male cocaine-dependent (DSM-IV) patients and in 28 controls of similar gender and age over a six month study. Exclusion criteria for the patients and controls included Axis I mood, anxiety or psychotic disorders. The patients could not manifest substance dependence except to cocaine. The patients and controls received remuneration for urine and blood collection. Blood samples for hormone levels were obtained between 8 and 10 a.m. on days 1, 14 and 21 of a 21-day inpatient treatment program and throughout 6 months of outpatient study visits at 45-day intervals. Attrition from treatment and study appointments occurred predominately at the junction between inpatient and outpatient programs. Forty percent of patients made the transition to outpatient treatment and remained abstinent and in treatment for a median of 103 days (ABST). Forty-two percent of patients dropped out of treatment during the inpatient stay or never returned after completing the inpatient program (DO) and 18% had a documented relapse either during, or within the first week after, the inpatient stay (REL). POMS total scores were elevated at treatment entry for both the ABST and DO groups. Plasma DHEAS levels in the DO patients were decreased compared to controls and increased in the ABST patients. POMS total scores for the REL patients at baseline were at control levels. Baseline cortisol levels were not statistically different between the outcome groups, though they were elevated for all cocaine patient groups. When treatment outcome was collapsed into whether patients completed (ABST) or did not complete 90 days of treatment (90N), ABST plasma DHEAS and cortisol were significantly elevated compared to the 90N patients and controls across the first 3 weeks of cocaine withdrawal. At treatment entry, each of the three patient outcome groups was identified by levels of circulating DHEAS and distressed mood. In the ABST patients, distressed mood during withdrawal may have been mitigated through antidepressant-like actions of enhanced endogenous DHEAS activity, thus contributing to improved abstinence and treatment retention. Patients, such as the DO group, with high levels of distressed mood at treatment entry and low DHEAS levels may benefit from adjunctive pharmacotherapy that targets DHEAS and POMS measures. Patients, such as the REL group, who lack distressed mood at treatment entry, may require intense application of motivational approaches plus residential treatment.