OBJECTIVES: Biochemical failure after definitive treatment for prostate cancer is common. We determined second treatment patterns in biochemical failure patients initially managed with radical prostatectomy (RP) and radiation (XRT). METHODS: Three hundred eighty-one and 259 patients managed with RP and XRT, respectively, who failed biochemically were identified from CaPSURE, a national prostate cancer registry. Failure was a PSA > 0.4 after RP and three consecutive rises in PSA above the nadir for XRT. Logistic regression was used to determine predictors of second treatment after failure. RESULTS: Median time to failure was 20.8 and 25.3 months for RP and XRT. One point twenty-one (32%) and 75 (29%) of RP and XRT failure patients received second treatment. Median time between failure and second treatment was 8.2 and 17.6 months for RP and XRT. Second treatments following RP were hormonal therapy (59%) and XRT (41%) and for XRT were mainly hormonal therapy (89%). Predictors of second treatment after biochemical failure for RP included PSA at diagnosis (p<.0001), T stage (p=.0062) and biopsy Gleason score (p=.0109). Patients with a PSA > 20, T3 disease and Gleason 8-10 had a 4.5 fold, 3.6 fold and 3.5 fold increased risk of second treatment as compared to patients with a PSA < 4, T1 disease, and Gleason 2-6, respectively. For XRT, predictors of second treatment included biopsy Gleason score (p<.0001) and PSA at diagnosis (p=.0057). Patients with Gleason 8-10 and PSA > 20 had a 5.4 fold and 6.5 fold increased risk of second treatment as compared to patients with Gleason 2-6 and PSA < 4, respectively. CONCLUSION: Second treatment timing and type differs for RP and XRT patients who fail biochemically. Pretreatment clinical factors can be used to determine which patients are at highest risk for second treatment. This has implications for initial treatment efficacy.