To review the biochemical relapse-free survival (bRFS) rates after treatment with permanent seed implantation (PI), external beam radiotherapy (EBRT) <72 Gy (EBRT <72), EBRT > or =72 Gy (EBRT > or =72), combined seeds and EBRT (COMB), or radical prostatectomy (RP) for clinical Stage T1-T2 localized prostate cancer treated between 1990 and 1998. The study population comprised 2991 consecutive patients treated at the Cleveland Clinic Foundation or Memorial Sloan Kettering at Mercy Medical Center. All cases had pretreatment prostate-specific antigen (iPSA) levels and biopsy Gleason scores (bGSs). Neoadjuvant androgen deprivation for < or =6 months was given in 622 cases (21%). No adjuvant therapy was given after local therapy. RP was used for 1034 patients (35%), EBRT <72 for 484 (16%), EBRT > or =72 for 301 (10%), PI for 950 (32%), and COMB for 222 patients (7%). The RP, EBRT <72, EBRT > or =72, and 154 PI patients were treated at Cleveland Clinic Foundation. The median radiation doses in EBRT <72 and EBRT > or =72 case was 68.4 and 78.0 Gy, respectively. The median follow-up time for all cases was 56 months (range 12-145). The median follow-up time for RP, EBRT <72, EBRT > or =72, PI, and COMB was 66, 75, 49, 47, and 46 months, respectively. Biochemical relapse was defined as PSA levels >0.2 for RP cases and three consecutive rising PSA levels (American Society for Therapeutic Radiology Oncology consensus definition) for all other cases. A multivariate analysis for factors affecting the bRFS rates was performed using the following variables: clinical T stage, iPSA, bGS, androgen deprivation, year of treatment, and treatment modality. The multivariate analysis was repeated excluding the EBRT <72 cases. The 5-year bRFS rate for RP, EBRT <72, EBRT > or =72, PI, and COMB was 81%, 51%, 81%, 83%, and 77%, respectively (p <0.001). The 7-year bRFS rate for RP, EBRT <72, EBRT > or =72, PI, and COMB was 76%, 48%, 81%, 75%, and 77%, respectively. Multivariate analysis, including all cases, showed iPSA (p <0.001), bGS (p <0.001), year of therapy (p <0.001), and treatment modality (p <0.001) to be independent predictors of relapse. Because EBRT <72 cases had distinctly worse outcomes, the analysis was repeated after excluding these cases to discern any differences among the other modalities. The multivariate analysis excluding the EBRT <72 cases revealed iPSA (p <0.001), bGS (p <0.001), and year of therapy (p = 0.001) to be the only independent predictors of relapse. Treatment modality (p = 0.95), clinical T stage (p = 0.09), and androgen deprivation (p = 0.56) were not independent predictors for failure. The biochemical failure rates were similar among PI, high-dose (> or =72 Gy) EBRT, COMB, and RP for localized prostate cancer. The outcomes were significantly worse for low-dose (<72 Gy) EBRT.