To characterize the effect of a prostate-rectum spacer on dose to rectum during external beam radiation therapy for prostate cancer and to assess for factors correlated with rectal dose reduction. Fifty-two patients at 4 institutions were enrolled into a prospective pilot clinical trial. Patients underwent baseline scans and then were injected with perirectal spacing hydrogel and rescanned. Intensity modulated radiation therapy plans were created on both scans for comparison. The objectives were to establish rates of creation of ≥ 7.5 mm of prostate-rectal separation, and decrease in rectal V70 of ≥ 25%. Multiple regression analysis was performed to evaluate the associations between preinjection and postinjection changes in rectal V70 and changes in plan conformity, rectal volume, bladder volume, bladder V70, planning target volume (PTV), and postinjection midgland separation, gel volume, gel thickness, length of PTV/gel contact, and gel left-to-right symmetry. Hydrogel resulted in ≥7.5-mm prostate-rectal separation in 95.8% of patients; 95.7% had decreased rectal V70 of ≥ 25%, with a mean reduction of 8.0 Gy. There were no significant differences in preinjection and postinjection prostate, PTV, rectal, and bladder volumes. Plan conformities were significantly different before versus after injection (P=.02); plans with worse conformity indexes after injection compared with before injection (n=13) still had improvements in rectal V70. In multiple regression analysis, greater postinjection reduction in V70 was associated with decreased relative postinjection plan conformity (P=.01). Reductions in V70 did not significantly vary by institution, despite significant interinstitutional variations in plan conformity. There were no significant relationships between reduction in V70 and the other characteristics analyzed. Injection of hydrogel into the prostate-rectal interface resulted in dose reductions to rectum for >90% of patients treated. Rectal sparing was statistically significant across a range of 10 to 75 Gy and was demonstrated within the presence of significant interinstitutional variability in plan conformity, target definitions, and injection results.