Purpose: Prostate cancer (PCa) is highly heritable. No validated PCa risk model currently exists. We therefore sought to develop a genetic risk model that can provide personalised predicted PCa risks based on known moderate-to-high-risk pathogenic variants, low-risk common genetic variants and explicit cancer family history, and to externally validate the model in an independent prospective cohort. Patients and methods: We developed a risk model using a kin-cohort comprising individuals from 16,633 PCa families ascertained in the UK in 1993-2017 from the UK Genetic Prostate Cancer Study, and complex segregation analysis adjusting for ascertainment. The model was externally validated in 170,850 unaffected men (7,624 incident PCas) recruited in 2006-2010 to the independent UK Biobank prospective cohort study. Results: The most parsimonious model included the effects of pathogenic variants in BRCA2, HOXB13 and BRCA1 and a polygenic score based on 268 common low-risk variants. Residual familial risk was modelled by a hypothetical recessively inherited variant and a polygenic component whose standard deviation decreased log-linearly with age. The model predicted familial risks that were consistent with those reported in previous observational studies. In the validation cohort, the model discriminated well between unaffected men and men with incident PCas within 5yr (C-index=0.790, 95% CI 0.783-0.797) and 10yr (C-index=0.772, 95% CI 0.768-0.777). The 50% of men with highest predicted risks captured 86.3% of PCa cases within 10yr. Conclusion: This is the first validated risk model offering personalised PCa risks. The model will assist in counselling men concerned about their risk and can facilitate future risk-stratified population screening approaches.