Background: Maintenance of physical function is a key consideration in treatment decision-making for older adults with metastatic cancer, many of whom are frail. However, physical function outcomes with treatment, and effects of frailty, have not been adequately explored in the mCRPC setting. We evaluated the effects of frailty status and treatment with docetaxel (CHEMO), abiraterone (ABI), enzalutamide (ENZA), and radium 223 (RAD) on elder-relevant physical function outcomes in older men with mCRPC. Methods: Men aged 65+ were enrolled in this multicenter prospective observational cohort study. Daily function was evaluated with the OARS instrumental activities of daily living (IADL). Objective physical function was assessed by grip strength and the Short Physical Performance Battery (SPPB). Falls were documented during interviews. We also collected FACT-G physical well-being (PWB) and functional well-being (FWB) subscales. Assessments were performed at baseline, 3 months, and 6 months. We identified frailty status with a validated frailty index. Mixed effects regression models were used to examine the difference in primary outcomes over time by treatment group or frailty status adjusted for baseline characteristics. Factors associated with falls within 6 months of treatment initiation were determined with logistic regression. Results: A total of 70, 38, 67, and 23 men starting CHEMO, ABI, ENZA, and RAD were included. Mean age, education, race, number of medications, and BMI were similar at baseline between treatment groups. In treatment-stratified analyses without considering frailty, no significant changes over time were reported for any physical function outcome. Frailty was significantly associated with lower IADL function (p < 0.0001), worse grip strength (p < 0.0001), worse SPPB score (p < 0.0001), worse PWB (p < 0.0001), and worse FWB (p < 0.0001) at baseline. In frailty-stratified analyses, grip strength (p = 0.0345) worsened, but SPPB (p = 0.0147) improved significantly over time. Also, changes in SPPB (p = 0.0394) and PWB scores (p = 0.0269) over time differed by frailty status, where frail cohorts had greater improvement over time in both scores. Frailty and treatment type were not predictors of falls whereas prior falls history (OR: 3.52, 95% CI: 1.40-8.86) and age (OR: 1.07, 95% CI: 1.01-1.14) were significant predictors. Conclusions: Frail older men receiving treatment for mCRPC have worse IADL function, grip strength, SPPB scores, PWB, and FWB at baseline. Although grip strength worsened over time, they had greater improvement in SPPB scores and PWB over time than fit patients. Contrary to our hypothesis, most older adults do not experience significant worsening in elder-relevant physical function outcomes over time regardless of treatment. The impact of frailty requires further study.