Tumor-bearing mice have a greater sensitivity to the acute lethal effects of the administration of high-dose recombinant human tumor necrosis factor alpha (rhTNF-alpha) compared to normal, non-tumor-bearing mice. We studied whether or not the presence of tumor per se was responsible for the enhanced rhTNF-alpha toxicity. Tumor-bearing mice underwent tumor excision or sham operation before the systemic administration of rhTNF alpha at staged times (0.5-24 h) following surgery. There was little survival difference between sham-operated tumor-bearing mice and tumor-bearing mice undergoing tumor excision (at 24 h, treatment with 12 micrograms rhTNF-alpha, survival:sham-operated tumor bearers = 0/12, excised tumor-bearers = 0/12; p2 less than 0.01 compared to non-tumor-bearers). Mice without tumors receiving sham operation, had minimal toxicity (10 of 12 mice surviving). The injection of 3 ml Ringer's lactate i.p. before i.v. rhTNF-alpha therapy increased survival in tumor-bearing animals; following pretreatment with Ringer's lactate 30/42 mice survived 12 micrograms rhTNF-alpha compared to 6/42 surviving a similar rhTNF-alpha dose without hydration (P2 less than 0.001). Since the production of oxygen free-radical metabolites has been postulated to play a role in the acute toxicity of rhTNF-alpha, bismuth subnitrate was used to induce the enzyme metallothionein to act as a natural scavenger for these metabolites. Daily oral bismuth subnitrate treatments improved survival of mice with MCA-106 or MCA-102 sarcoma and of mice without tumors, with higher rhTNF-alpha doses (12-20 micrograms), without reducing the therapeutic effect of rhTNF-alpha against the weakly immunogenic MCA-106 sarcoma. These studies suggest methods for reducing the toxicity of rhTNF-alpha administration in clinical trials.