Survival in a group of 556 patients with uveal melanoma treated by proton beam irradiation with a median follow-up of 5.3 years was compared with that of 238 patients enucleated during the same 10-year period as irradiated patients (July 1975 to December 1984) with a median follow-up of 8.8 years, and 257 patients enucleated during the preceding 10 years (January 1965 to June 1975) with a median follow-up of 17.0 years. Adjustments were made for known prognostic factors including age, tumor location, tumor height, and clinical estimate of tumor diameter (for enucleated patients this was estimated in a regression equation relating histologic to clinical measurement). The overall rate ratio for all cause mortality was 1.2 (95% confidence interval, 0.9-1.6) for the concurrent enucleation series versus proton beam, and 1.6 (95% confidence interval, 1.2-2.1) for the earlier enucleation series versus proton beam. Relative rates of metastatic death, cancer death, and all cause mortality comparing alternative treatments were found to vary with time after treatment. Interval-specific rate ratios were evaluated using proportional hazards models fitted to separate time intervals after treatment. For all three outcomes, rate ratios were over two and statistically significant for the first 2 years after treatment and closer to one and nonsignificant after year 6 comparing the two enucleation groups with proton beam. Results suggest that treatment choice has little overall influence on survival in patients with uveal melanoma.