We compared the survival of 302 patients with a primary choroidal or ciliary body melanoma treated by cobalt-60 plaque radiotherapy between 1976 and 1982 with the survival of 134 patients treated by enucleation during the same period. Tumor size, location of the anterior margin of the tumor, and patient age at the time of treatment were identified as simultaneous significant clinical variables for predicting melanoma-specific mortality by multivariate Cox proportional hazards modeling. We computed a prognostic index for each patient based on this model and found that patients in the enucleation group had slightly higher values of this index than did patients in the cobalt-60 plaque radiotherapy group. Risk ratios for the treatment effect computed from a Cox model incorporating prognostic index and the treatment variable were found to be approximately equal to 1, both for analysis of melanoma-specific mortality and total mortality. These results indicate that when one controls for differences in prognostic index between the groups, cobalt-60 plaque therapy and enucleation are essentially equivalent in their effect on survival.