A total of 80 patients with melanoma metastases in regional lymph nodes were treated by i.d. injections with a vaccine prepared from a vaccinia virus-infected allogeneic melanoma cell line; 39 patients have been followed for a 2-year period. Interim results from comparison of the treated group with 151 historical controls treated without the vaccine from September 1978 to December 1981 at the same institution and 56 non-randomized concurrent controls suggest that survival was significantly prolonged in the vaccinated group. At the 2-year period overall survival was 75% in the treated compared to 57% in the historical control group. Subset analysis showed a greater apparent benefit of vaccine therapy among patients who had metastases detected at the time of treatment of the primary melanoma (synchronous metastases), while therapy appeared less effective in patients with metastases detected at some time after treatment of the primary (delayed metastases). In the latter only those with one lymph node appeared to benefit from the, treatment whereas in patients with synchronous metastases patients with three or more nodes as well as one node appeared to have improved survival. The survival rates at 2 years for treated patients with synchronous metastases in one, two, three or more lymph nodes was 100%, 83% and 79% respectively compared with that of 82%, 86% and 47% respectively in the equivalent control groups. Survival rates in treated patients with delayed metastases in one, two, three or more lymph nodes was 70%, 70% and 65% compared with 47%, 42% and 35% in the equivalent control groups. Treatment and control groups appeared well matched for a number of known prognostic features, including number and size of involved nodes, sex and thickness of primary tumor. Multivariate analysis indicated the effect of treatment was independent of these factors. Despite the empiricism of this approach the present results suggest that this form of therapy warrants further evaluation in a randomized controlled trial.