Metabolic abnormalities in obesity and non-insulindependent diabetes mellitus (NIDDM) are linked to insulin resistance [1±4]. Although the mechanism for the development of insulin resistance has not been fully clarified, recent studies imply a correlation between tumour necrosis factor-a (TNF-a) and insulin resistance. These studies demonstrated that TNF-a mRNA expression is increased in adipose tissue of obese animals [4±6] and humans [7,8], that infusion of TNF-a into rats results in insulin resistance in muscle [9] and that neutralization of TNF-a by infusion of TNF-a receptor-IgG chimeric protein into obese rats increases peripheral glucose uptake in response to insulin [4]. It was also shown that TNF-a mRNA is overexpressed in muscle of diabetic patients [10] and that the RNA transcript encoding the p55 TNF-a receptor is elevated in muscle of genetically obese and diabetic KKA y mice [11]. Since the half-life of TNF-a in blood is very short and muscle plays a principal role in inducing impaired glucose metabolism in insulin resistance [12], it is supposed that TNF-a produced in muscle acts in an autocrine fashion or that produced in adipose tissue works on the adjoining muscle. Diabetologia (1998) Summary In order to evaluate the relationship between tumour necrosis factor-a (TNF-a) level in muscle and metabolic abnormalities in obesity and diabetes mellitus, pioglitazone, a novel insulin-sensitizing agent, was administered to Wistar fatty rats and time-dependent changes in muscle TNF-a content and plasma indicators of diabetes and obesity were measured. Wistar fatty rats were hyperglycaemic, hyperlipidaemic and hyperinsulinaemic, and their plasma and muscle TNF-a levels were two or more times higher than those in normal lean rats at 16 weeks of age. When pioglitazone was administered to fatty rats at a dose of 3 mg × kg ±1 × day ±1 , the plasma triglyceride level and TNF-a levels in plasma and muscle decreased time-dependently, and reached the levels of lean rats within 4 days. Plasma glucose and insulin levels also decreased time-dependently with pioglitazone, but on day 4, these levels were still much higher than the levels in lean rats. Neutral sphingomyelinase (SMase) activity in muscle of fatty rats was two times higher than that in lean rats and was lowered to the level of that in lean rats by 4 days' pioglitazone administration. The plasma leptin level in fatty rats was 8 times higher than that in lean rats, but pioglitazone did not affect the level during the 4-day administration period. These results suggest that an increase in TNF-a production and subsequent activation of SMase in muscle leads to metabolic abnormalities in obesity and diabetes and that antidiabetic activity of pioglitazone is deeply associated with the suppression of TNF-a production. [Diabetologia (1998) 41: 257±264]