We conducted a phase II trial of the novel immunomodulatory/cytotoxic agent 15-deoxyspergualin in patients with metastatic breast cancer who had failed treatment with front-line chemotherapy. Thirty-eight courses of treatment were administered to fourteen patients enrolled in this trial, 25 at a dose of 1800 mg/m2/d (dose level 0) and 13 at a dose of 2150 mg/m2/d (dose level +1) administered by continuous intravenous infusion for 5 days. Treatment was well tolerated with neuromuscular side-effects (myalgias, paresthesias) and granulocytopenia (nadir granulocyte count of 0.50-0.99 x 10(9)/l) in two and three courses, respectively, as the only grade III toxicities. The neuromuscular toxicity of deoxyspergualin is probably related to the occurrence of hypomagnesemia. No partial or complete responses were observed in this study. One patient achieved a minor response but had progressive disease 65 weeks after enrollment. The response was observed coincident with an increase in T4/T8 ratio in the peripheral blood. The median time to progression for the entire cohort was eight weeks (range, 4-65 weeks). There was no clinical evidence of immunosuppression and no decrease in total peripheral blood lymphocyte counts or helper T-cells was observed. At the doses and schedule employed in this trial, deoxyspergualin does not appear to have significant activity against metastatic breast cancer resistant to front-line chemotherapy. The correlation between hypomagnesemia and neuromuscular toxicity of deoxyspergualin is an intriguing, previously unknown observation and requires further investigation.