While additive in vitro antitumor cytotoxicity has been observed with tumor necrosis factor (TNF) in combination with cytotoxic agents, particularly actinomycin D (AMD), the critical issue of potentially enhanced host toxicity has not been assessed with such combinations. Dose- and schedule-dependent toxicity of AMD and recombinant human TNF (rHTNF) were studied in mice using sublethal single agent doses. Dose-dependent toxicity was tested using 500, 400, 250, and 50 micrograms/kg of AMD (i.p.) and 50, 25, and 5 micrograms/kg of rHTNF (i.v.). AMD was injected 2 h before TNF. The lethality from maximal doses of these agents alone was consistent with the LD10. However, in combination, an 83% (40/48) lethality was observed at the 500 micrograms/kg dose of AMD, and varying the TNF dose tenfold had no pronounced impact on mortality. A TNF dose response was observed with 400 micrograms/kg of AMD. No deaths occurred with the lower AMD doses (regardless of TNF dose). A similar pattern of lethality at the same doses of AMD and TNF was seen in immunocompetent heterozygotes from the same inbred strain. Toxicity was critically dose schedule (order and interval) dependent. Substantial lethality was seen only if AMD preceded TNF and was maximal at a dose interval of 15 min; this pattern was seen at two dose ranges. If AMD was given after TNF, essentially no lethality was seen. If Phase I clinical trials of AMD/TNF are to be contemplated, the design of such protocols must address (a) the likelihood that severe toxicity might occur at doses far lower than with rHTNF alone, (b) the importance of dose schedule on toxicity, and (c) the critical necessity for precise dose administration compliance if consistent results are to be achieved.