We conducted a phase II trial of bleomycin+vincristine+lomustine+dacarbazine (BOLD) with intercycle alpha interferon-2b in previously untreated patients with metastatic uveal melanoma. Objective tumor response and toxicity were assessed. Twenty-three patients with histologically verified metastatic uveal melanoma were enrolled into this study between November 1992 and August 1995. Chemotherapy was administered in the following fashion: dacarbazine (DTIC), 200 mg/m2 intravenously on days 1-5; vincristine, 1 mg/m2 (not to exceed 2 mg) intravenously on days 1 and 4; bleomycin, 15 mg intravenously on days 2 and 5; lomustine (CCNU), 80 mg orally on day 1; and alpha interferon-2b, 3 x 10(6) IU subcutaneously on days 8, 10, 12, 15, 17, 19. A cycle was 28 days, and patients were reevaluated after every 2 cycles. Among twenty evaluable patients, four objective responses were observed (RR = 20%). Hematologic toxicity was modest by comparison to some other combination chemotherapy regimens in common use. Neurotoxicity was frequently observed, but it was seldom severe. An unexpected and unpredictable severe pulmonary toxicity was observed in 3 patients, the etiology of which remains unclear. The regimen of BOLD+interferon is active in the treatment of metastatic uveal melanoma. The precise role of the regimen has to be defined in light of its toxicity, particularly the unpredictable pulmonary toxicity. The pattern of occurrence of these pulmonary events is most consistent with either an acquired hypersensitivity reaction or a cumulative toxic effect of 2 or more of the agents. Patients considered for treatment with this regimen must be judiciously selected. Those with no clear contraindications may benefit from a trial of this regimen, but they must be monitored closely.