We investigated the capacity of several recombinant cytokines to induce IL-6 in vivo in both normal and tumor-bearing (TB) mice. Intravenous administration of human rhTNF-alpha, rhIL-1, rhIL-2, rhIFN-alpha A/D, and rmIFN-gamma were all capable of inducing circulating IL-6. rhTNF-alpha administration caused the greatest induction of IL-6. TB animals consistently produced more IL-6 in response to rhTNF-alpha than did normal mice (2 h after 4 micrograms rhTNF-alpha, TB = 24,100 HGF U/ml, non-TB = 3600 HGF U/ml of IL-6). A single daily i.v. dose of rhTNF-alpha (4 micrograms/mouse/day) for 5 days led to decreased IL-6 induction in TB animals by day 3 of treatment (peak levels of IL-6, day 1 = 72,800 HGF U/ml, day 3 = 23,400 HGF U/ml, day 5 = 26,400 HGF U/ml). rhIL-1 administration also resulted in considerable IL-6 production, although peak values were less than those resulting from administration of rhTNF-alpha. Administration of rhIL-1 induced similar IL-6 levels (TB = 10,025 and non-TB = 10,600 HGF U/ml) in TB and normal mice. Single high doses of rhIL-2, rhIFN-alpha A/D, and rmIFN-gamma induced lower but consistent levels of circulating IL-6 in mice with and without tumor. In addition, the sera of untreated TB mice contained levels of IL-6 which paralleled the extent of tumor burden (serum IL-6 in day 30 MCA 106 TB mice = 420 HGF U/ml). The detection of de novo IL-6 was also confirmed in animals bearing tumors of different histologies (the MCA 102 sarcoma, MCA 38 adenocarcinoma, and B16 melanoma). At no time was IL-6 measurable in the sera of untreated normal mice. The identification of IL-6 was verified by neutralization studies using specific antimurine IL-6 antibody. Although the exact role of IL-6 in TB animals remains to be elucidated, its known pleotrophic immune and metabolic effects may be important in the host response to malignancy.