Purpose: Adrenal androgens activate the androgen receptor and stimulate prostate cancer growth. Ketoconazole is used as an inhibitor of adrenal androgen synthesis in men with androgen-independent prostate cancer. This study analyzes the relationship between pretreatment androgen levels and outcome following ketoconazole treatment. Experimental Design: Baseline levels of three adrenal androgens (androstenedione, dehydroepiandrostenedione, and dehydroepiandrostenedione-sulfate) and testosterone were measured. Regression models (logistic and proportional hazard) were used to assess the prognostic significance of these levels in predicting overall survival and prostate-specific antigen (PSA) response defined by Consensus Criteria. Results: In 103 patients with available levels, PSA response rate was 28% and median response duration was 4.8 months. The median baseline androstenedione level was 0.64 ng/mL and was 0.88 ng/mL versus 0.53 ng/mL for those with and without a PSA response, respectively (P = 0.034). In univariate analysis, elevation of baseline androstenedione levels was predictive of PSA response [odds ratio, 2.26; 95% confidence interval (95% CI), 1.03-4.96; P = 0.043]. In multivariate analysis, both the uppermost and the middle tertile of baseline androstenedione level were associated with an improved overall survival compared with those in the lower tertile (hazard ratio, 0.59; 95% CI, 0.36-0.98; P = 0.40; hazard ratio, 0.53; 95% CI, 0.32-0.90; P = 0.018, respectively). A linear correlation was observed among all androgen levels. Conclusions: Higher androstenedione levels predict likelihood of response to ketoconazole and improved survival compared with patients with lower levels. These data suggest that therapy with ketoconazole is less effective in patients with low levels of androgen at baseline.