BCR is the most clinically used endpoint for identification of treatment failure. Approximately 15% to 53% of patients undergoing primary curative therapy will develop BCR. BCR often precedes clinically detectable recurrence by years. It does not necessarily translate directly into PCa morbidity and mortality, nor does it always reflect the desired endpoint. Furthermore, it has not been validated as a surrogate endpoint, in that interventions that have been shown to alter the PSA level have not been shown to also alter survival. The utility of PSA level as a surrogate endpoint is brought into question by the knowledge that the overall survival rate of patients at 10 years is similar in patients with and without BCR, and that in a significant proportion of men, the only evidence of disease during their lifetime will be a detectable PSA level. The likelihood of developing BCR post-therapy can be predicted by using multiple clinical and pathologic variables. With the development of nomograms that incorporate several markers, the accuracy of prediction has improved. Until recently, the natural history of BCR post-RRP has not been well understood. Pound et al showed the heterogenous and prolonged natural history of BCR. In this large series of men with BCR following RRP, only 34% of men developed metastatic disease. The median time from development of BCR to identification of metastases was 8 years, and the median time from the development of metastatic disease to death was just under 5 years. These data highlight the extremely variable and potentially indolent nature of BCR. The risk of metastatic disease following BCR has been relatively well defined and relates to PSADT and time to PSA recurrence. It generally is accepted that a PSADT of less than 6 to 10 months and a time to PSA recurrence of less than 1 to 2 years relates to a higher risk of developing metastatic disease. Local recurrence, however, remains poorly understood with respect to its true incidence, clinical significance, and natural history. The significance of BCR post-RT remains unclear due to the lack of data on its natural history. Attempts have been made to identify patients at high risk for metastatic progression by looking at time to PSA recurrence and PSADT. A PSADT of less than 6 to 12 months and a time to PSA recurrence of less than 12 months reflects a higher risk of developing metastatic disease. Accurate risk stratification by means of an algorithm similar to that produced by Pound et al has not been performed on a large cohort, thus making risk assessment for an individual patient difficult. The major dilemma for clinicians in the management of BCR is the identification of the site of disease recurrence, which ultimately guides therapy decisions. Clinicopathologic features allow for risk stratification for recurrence, and multiple investigations have attempted to localize the site of recurrence. Time to biochemical progression, Gleason score, and PSADT are predictive of the probability and time to development of metastatic disease, and allow for stratification of patients into different risk groups (see Table 2). TRUS, CT, PET, and DRE all have limited utility in the identification of local recurrence. ProstaScint and MRI have demonstrated encouraging initial results: however, they require further investigation. Bone scintigraphy is of little value for the initial investigation of BCR. In patients with a PSA level of less than 10 ng/mL, the risk of having a positive bone scan is less than 1% and, until the PSA level rises above 40 ng/mL, the risk of having a positive bone scan is less than 5%. Therefore, bone scintigraphy should be reserved for patients with a PSA level greater than 10 to 20 ng/mL or patients with a rapidly rising PSA level. Using new MRI sequences, there is some evidence that MRI is better for the detection of bony metastatic disease; however, this technique requires further investigation. BCR causes anxiety for the patient and the treating doctor, because the best way to manage patients with PSA-only progression is unknown. Currently, there are no validated treatment recommendations for the management of BCR. The information in this review provides the framework for assignment of patients into clinical trials based on different risk categories. Patients at high risk for metastatic progression could be identified early and thus entered into appropriate clinical trials for systemic therapies. Similarly, patients with a low risk of progression could be placed into observation protocols, potentially sparing them from exhaustive and inappropriate investigations.