Purpose Biochemical failure (BF), defined by a rising prostate-specific antigen (PSA) profile, is an early surrogate of treatment failure. However, little evidence is available to show that BF is associated with death for patents with prostate cancer treated with radiotherapy. We examined the relationship between BF and death from prostate cancer. Methods and materials A total of 942 patients were treated between 1987 and 1998 with external beam radiotherapy who had sufficient PSA determinations in follow-up for the analyses described. The median radiation dose was 72 Gy, median PSA was 9.9 ng/mL, and median follow-up was 73 months. The American Society for Therapeutic Radiology and Oncology consensus definition was used to define BF. Kaplan-Meier calculations were from the start of radiotherapy. Cox proportional hazards regression multivariate analyses were used to investigate the association of BF (time-dependent variable) and other factors to distant metastasis (DM), cause-specific death (CSD), and overall death (OD). The year of treatment was included in some of the multivariate analyses to correct for potential unknown factors that may have occurred during the years of the study, such as stage migration. Results BF was observed in 316 patients (34%), and 66 (7%) experienced DM, 32 (3%) died of prostate cancer, and 230 (24%) died overall during the study period. The Kaplan-Meier 5-year rate estimates from the start of treatment for BF, DM, CSD, and OD were 38%, 6%, 3%, and 13%, respectively. All patients with DM had BF. In multivariate analyses, BF was associated with DM and CSD, but not OD. The inclusion of the year of treatment did not alter these relationships. Conclusion BF, as a time-dependent covariate, was the strongest determinant of DM and was also very significantly related to CSD. The inclusion of the year of treatment had little effect on these associations. Longer follow-up is needed to determine conclusively the relationship of BF to OD.