Background: Currently,overall survival (OS) is used to determine treatment efficacy in trials for advanced prostate cancer (PC). Reliable intermediate endpoints could hasten drug development efforts. We therefore sought to determine the association between an intermediate endpoint—time to metastasis (TTM)—and OS in PC patients receiving androgen deprivation therapy (ADT) for biochemical recurrence (BCR). Methods: The Dana-Farber PC database identified 415 patients who received ADT for non-metastatic BCR with median 6.4 years follow-up. Associations between TTM and OS were measured from (i) time of ADT initiation (n=398) and (ii) time of castration-resistance (n=247) using a non-parametric Kendall tau rank correlation for bivariate time to event outcomes. A Cox regression model was used to assess the association of development of metastasis and OS for landmark timepoints. Results: Among 398 subjects analyzed from start of ADT, 180 (45%) developed metastases with median TTM 6.25 years. A total of 152 (38%) died, with a median OS of 8.5 years. For men developing metastases within five years of ADT initiation, hazard ratio (HR) for mortality was 7.42 (p<0.0001, 95% CI 4.18-13.17). Among 247 subjects assessed from time of castration resistant PC (CRPC), 172 (70%) developed metastases with a median TTM 2.6 years. In this subgroup, 140 (56%) patients died with median OS 5.25 years. For men developing metastases within three years of developing castration-resistance, HR for mortality was 5.25 (p<0.0001, 95% CI 3.20-8.61). Overall, correlation between TTM and OS from ADT initiation and OS was 0.25 (P<0.0001, 95% CI 0.20-0.29). Correlation between TTM and OS from CRPC was 0.32 (P<0.0001, 95% CI 0.25-0.37). Conclusions: TTM was significantly correlated with OS in PC patients treated with ADT for BCR. Prospective data in a large validation cohort will help determine whether TTM is suitably predictive of OS to serve as a primary endpoint in clinical trials.