To prospectively develop a prostate cancer (CaP) risk calculator in a racially diverse population. All patients referred for prostate biopsy due to an elevated prostate-specific antigen or abnormal digital rectal exam in a 19-months period at Kaiser Permanente Northern California underwent a standardized systematic, ultrasound-guided biopsy scheme (14-cores for initial biopsy, 18-20 cores for repeat biopsy). All pertinent clinical variables were prospectively collected. The highest Gleason score for each patient was recorded for all positive biopsies. We used a split sample design to develop and validate 3 multivariable prediction models using multinomial logistic regression with the least absolute shrinkage and selection operator. All models included these core variables: age, race, prostate-specific antigen, prior biopsy status, body mass index, and family history of CaP. Model 1 included only the core variables, Model 2 added digital rectal exam, and Model 3 added digital rectal exam and prostate volume. We considered 3 outcomes: high-grade disease (Gleason score ≥7), low-grade disease (Gleason score = 6), and no cancer. Predictive discrimination was quantified using the c-statistic. Complete data were available for 2,967 patients. Cancer was found in 50% of patients: of these, 58% were Gleason score ≥7 and 42% were low grade. Compared to Caucasians, African Americans were at a higher risk while Asians and Hispanics were at a lower risk for overall and high-grade cancer detection. The number of prior negative biopsies was also protective for these outcomes. The c-statistics for Model 1, 2, and 3 to predict high-grade disease vs. low-grade or no cancer were 0.76, 0.79, and 0.85, respectively. The c-statistics for Model 1, 2, and 3 to predict any CaP vs. no cancer were 0.69, 0.70, and 0.79, respectively. All models were well calibrated for all outcomes. In men with elevated PSA levels, our calculator provides useful information that may enhance the shared decision-making process regarding the need for biopsy.