Abstract Background Patients with castration-resistant prostate cancer (CRPC) and bone metastases have an unmet clinical need for effective treatments that improve quality of life and survival with a favorable safety profile. Objective To prospectively evaluate the efficacy and safety of three different doses of radium chloride (Ra 223) in patients with CRPC and bone metastases. Design, setting, and participants In this phase 2 double-blind multicenter study, 122 patients were randomized to receive three injections of Ra 223 at 6-wk intervals, at doses of 25 kBq/kg (n=41), 50 kBq/kg (n=39), or 80 kBq/kg (n=42). The study compared the proportion of patients in each dose group who had a confirmed decrease of ≥50% in baseline prostate-specific antigen (PSA) levels. Outcome measurements and statistical analysis Efficacy was evaluated using blood samples to measure PSA and other tumor markers, recorded skeletal-related events, and pain assessments. Safety was evaluated using adverse events (AEs), physical examination, and clinical laboratory tests. The Jonckheere-Terpstra test assessed trends between groups. Results and limitations The study met its primary end point with a statistically significant dose–response relationship in confirmed ≥50% PSA declines for no patients (0%) in the 25-kBq/kg dose group, two patients (6%) in the 50-kBq/kg dose group, and five patients (13%) in the 80-kBq/kg dose group (p=0.0297). A ≥50% decrease in bone alkaline phosphatase levels was identified in six patients (16%), 24 patients (67%), and 25 patients (66%) in the 25-, 50-, and 80-kBq/kg dose groups, respectively (p<0.0001). The most common treatment-related AEs (≥10%) occurring up to week 24 across all dose groups were diarrhea (21%), nausea (16%), and anemia (14%). No difference in incidence of hematologic events was seen among dose groups. Potential limitations include small patient numbers and differences among dose groups at baseline. Conclusions Ra 223 had a dose-dependent effect on serum markers of CRPC activity, suggesting that control of bone disease with Ra 223 may affect cancer-related outcomes. Ra 223 was well tolerated at all doses. Trial registration ClinicalTrials.gov: NCT00337155.