Background: In non-metastatic CRPC with short PSA doubling time, APA or ENZA with ADT have shown to slow development of metastasis. We performed an indirect comparison of APA and ENZA clinical effectiveness (efficacy and side effects) using network meta-analysis. Methods: A search of Medline, Embase, Cochrane Library, and Cochrane Central Register of Controlled Trials was performed to identify relevant RCTs. Collected data included hazard ratio (HR) and confidence interval (CI) for primary outcome of median time to metastasis free survival (MFS) and secondary outcomes of median time to overall survival (OS), PSA progression, initiation of cytotoxic therapy and number of adverse events (AE) in each study arm. Risk for bias was assessed using the Cochrane Collaboration’s tool. We used a fixed effects model using Bayesian approach to generate pooled estimates. Results: Two randomized trials (n=2608) were analyzed. SPARTAN trial compared APA plus ADT to ADT alone and PROSPER compared ENZA plus ADT to ADT alone. Risk of bias was low. There were no statistically significant differences between APA and ENZA in MFS, OS, median time to PSA progression and initiation of cytotoxic therapy. The MFS endpoint hazard ratio (HR) was 0.97 (95% CI 0.73-1.28) for comparing the two drugs. APA had fewer reports of hypertension (HR 0.54, 95% CI 0.31–0.91) compared to enzalutamide. However, no significant differences were observed in other AEs including grade 3 or higher AE, serious AE, AE leading to death or trial discontinuation. Common AEs (nausea, diarrhea, fall, dizziness, arthralgia, weight loss) and AEs of special interests (mental impairment disorders, seizures) were also comparable between APA and ENZA. Conclusions: In non-metastatic CRPC with short PSA doubling time, APA and ENZA showed similar time to metastasis free survival. APA was associated with lower incidence of hypertension than ENZA; otherwise, side effect profile was similar. Given similar efficacy and risk profile demonstrated in this study, clinical decision making should depend on other factors such as cost and availability.