purpose: The pleiotropic inflammatory effects of tumor necrosis factor (TNF) prompted a study of this cytokine in familial Mediterranean fever (FMF), a recurrent polyserositis of unknown etiology. patients and methods: Thirty-six asymptomatic and 24 patients with acute FMF were studied and compared with 20 matched healthy subjects. TNF levels were measured by bioassay in the plasma and in supernatants of peripheral blood mononuclear cells (PBMC) incubated alone or with an inducer (lipopolysaccharide, phytohemagglutinin [PHA], or Sendai virus). Cytotoxicity could be abolished in all cases by preincubation with monoclonal anti-TNF-α antibodies. results: No TNF was found in plasma and non-induced PBMC supernatants. Induced TNF production was markedly decreased in patients with acute FMF and increased in asymptomatic FMF patients to levels over those of control subjects (p <0.05). Thus, PHA-induced TNF levels were 4 U/mL in patients with acute FMF, 25 U/ mL in asymptomatic patients, and 14 U/mL in healthy control subjects (median values), and the other inducers gave similar results. Retesting of patients first studied during an acute episode when their disease was quiescent also revealed a fivefold increase in TNF production. These results were independent of the use of colchicine, which also had no effect on TNF levels when taken by volunteers (1 mg/day) or when added to the PBMC cultures (10-7 M). conclusions: Since TNF has a very short half-life in plasma, the capacity of PBMC to respond to TNF inducers may more accurately reflect its synthesis. A marked decrease in this response in acute FMF suggests "exhaustion" of cells that are already highly activated to produce TNF and the possible participation of TNF in the pathogenesis of FMF.