Natural killer cells spontaneously lyse certain tumor cells and may defend against malignancy. We have previously shown that natural killing (NK) by human peripheral blood mononuclear cells (PBMC) is suppressed in vitro by phorbol diester tumor promoters, including 12-O-tetradecanoylphorbol-13-acetate (TPA). We here demonstrate that suppression of NK is mediated by monocytes or polymorphonuclear leukocytes (PMN) and that suppression is dependent on the generation of reactive forms of molecular oxygen (RO), particularly hydrogen peroxide (H2O2). NK was suppressed not only by TPA but also by opsonized zymosan (yeast cell walls), which, like TPA, was not toxic to PBMC. Both TPA and zymosan stimulated the production of superoxide anion (O2-) and H2O2 by PBMC. Production of RO correlated with suppression of NK. When PBMC were depleted of monocytes, the production of RO and the suppression of NK were both markedly reduced. Suppression could be restored by monocytes or PMN, both of which produced RO in response to TPA or zymosan. Suppression of NK was dependent on RO. Monocytes or PMN from a patient with chronic granulomatous disease, whose cells cannot generate RO, did not mediate suppression of NK. Suppression was also reduced in glucose-free medium, which did not support the generation of RO. Suppression of NK by TPA was inhibited by catalase. Bovine superoxide dismutase had a limited effect on suppression, even in high concentration, and tyrosine-copper (II) complex, which also enhances dismutation of O2- to H2O2, had almost no effect on suppression. When H2O2 was directly generated enzymatically from glucose oxidase and glucose, NK was suppressed and suppression was reversed by catalase. NK was also suppressed by the enzymatic generation of O2- from xanthine oxidase and xanthine, but suppression under these conditions was again inhibited by catalase and not by superoxide dismutase, indicating that suppression was due to the secondary formation of H2O2 from O2-. These results indicate that H2O2 is important in suppression of NK. Myeloperoxidase did not appear to play a role in suppression because inhibition of this enzyme by sodium azide, cyanide, or aminotriazole did not prevent suppression of NK. Suppression of NK was reversible; after exposure to zymosan, NK could be partially restored by the addition of catalase and superoxide dismutase or by the removal of zymosan. These studies demonstrate cellular regulation of NK by monocytes or polymorphonuclear leukocytes and indicate a role for RO in immunoregulation.