BACKGROUND 125I episcleral plaque therapy has gained wide acceptance for the treatment of uveal melanoma because of its potential to preserve vision, salvage the globe, and provide good local control. A rigorous analysis of the optimum radiation dose, dose rate, and overall treatment time has not been reported with this technique. METHODS One hundred fifty patients with uveal melanoma treated with 125I plaques between 1982 and 1990 and included in the uveal melanoma study (UMS) database of the Wills Eye Hospital were analyzed. Mean patient age was 60.7 years (range: 17.7-84.6 years). Initial mean tumor size was 9.7 x 8.5 x 3.7 mm with a range of 4.5 to 21.5 mm in basal dimension and 1.2 to 11.8 mm in height. Mean dose to the tumor apex was 94.77 gray (Gy) (29.5-141 Gy). Mean dose rate to the tumor apex was 92.9 cGy/hr (10-292 cGy/hr); the mean dose to the base was 359 Gy (181-692 Gy); the mean dose rate to the base was 348 cGy/hr (112-893 cGy/hr); and mean duration of treatment was 124.7 hours (range: 28-333 hours) RESULTS With a median follow-up of 68 months, there have been 33 local failures. Mean time to local failure was 19 months (range: 6-78 months). Actuarial local control is 81% at 5 years. Multivariate analysis demonstrates significant correlation of local failure with larger tumor dimension (P = 0.0046), close proximity to the optic disc (P = 0.0029), lower radiation dose to the tumor apex (P = 0.03), lower radiation dose rate to the tumor apex and base (P = 0.01 and 0.03), and longer overall treatment time (P = < 0.0001). CONCLUSIONS This retrospective analysis reinforces the importance of dose rate, minimum tumor dose, overall treatment time, maximum tumor basal dimension, and proximity to the optic nerve in the treatment of uveal melanoma.