To identify dosimetric, anatomic, and clinical factors that correlate with late rectal toxicity after three-dimensional conformal radiotherapy (3D-CRT) for prostate cancer. We retrospectively analyzed the dose-volume histograms and clinical records of 163 Stage T1b-T3c prostate cancer patients treated between 1992 and 1999 with 3D-CRT, to a total isocenter dose of 74-78 Gy at The University of Texas M. D. Anderson Cancer Center. The median follow-up was 62 months (range 24-102). All late rectal complications were scored using modified Radiation Therapy Oncology Group and Late Effects Normal Tissue Task Force criteria. The 6-year toxicity rate was assessed using Kaplan-Meier analysis and the log-rank test. A univariate proportional hazards regression model was used to test the correlation between Grade 2 or higher toxicity and the dosimetric, anatomic, and clinical factors. In a multivariate regression model, clinical factors were added to the dosimetric and anatomic variables to determine whether they significantly altered the risk of developing late toxicity. At 6 years, the rate of developing Grade 2 or higher late rectal toxicity was 25%. A significant volume effect was observed at rectal doses of 60, 70, 75.6, and 78 Gy, and the risk of developing rectal complications increased exponentially as greater volumes were irradiated. Although the percentage of rectal volume treated correlated significantly with the incidence of rectal complications at all dose levels (p <0.0001 for all comparisons), the absolute rectal volume appeared to be a factor only at the higher doses of 70, 75.6, and 78 Gy (p = 0.0514, 0.0016, and 0.0021, respectively). The following variables also correlated with toxicity on the univariate analysis: maximal dose to the clinical target volume, maximal dose to rectum, maximal dose to the rectum as a percentage of the prescribed dose, and maximal dose delivered to 10 cm(3) of the rectum. Of the clinical variables tested, only a history of hemorrhoids correlated with rectal toxicity (p = 0.003). Multivariate analysis showed that the addition of hemorrhoids increased the risk of toxicity for each dosimetric variable found to be significant on univariate analysis (p <0.05 for all comparisons). Dose-volume histogram analyses clearly indicated a volume effect on the probability of developing late rectal complications. Therefore, dose escalation may be safely achieved by adherence to dose-volume histogram constraints during treatment planning and organ localization at the time of treatment to ensure consistent patient setup.