The effects of troglitazone, a novel thiazolidinedione, in non-insulin-dependent diabetic (NIDDM) patients were studied in a double-blind, parallel-group, placebo-controlled, dose-ranging trial. A total of 330 patients (63 % male), mean age 57 years (range 39-72), with two fasting capillary blood glucose values _> 7 and ___ 15 retool/1 (within 2.5 mmol/1 of each other) were randomised to treatment with placebo or troglitazone at doses of 200, 400, 600 or 800 mg once daily, or 200 or 400 mg twice daily, for 12 weeks. Prior to the study, treatment had been with diet alone (38 % patients) or with oral hypoglycaemic agents which were stopped 3-4 weeks before study treatment started. During treatment, HbA1 c tended to rise in patients taking placebo (7.2-8.0%), but remained unchanged with all doses of troglitazone. After 12weeks of treatment, HbAlc was significantly lower in the troglitazone-treated (mean 7.0-7.4%) compared to the placebo-treated (8.0 %) patients (p --0.055 to < 0.001), as was fasting serum glucose concentration (troglitazone, 9.3-11.0 mmol/1 vs placebo, 12.9 mmol/1, p < 0.001). All doses of troglitazone were equally effective. Troglitazone also lowered fasting plasma insulin concentration, by 12-26 % compared to placebo (p = 0.074 to < 0.001). Insulin sensitivity assessed by homeostasis model assessment (HOMA) was greater after 12 weeks of treatment in troglitazone-treated patients (troglitazone, 34.3-42.8 % vs placebo, 29.9 %, p < 0.05). In addition, serum triglyceride and non-esterified fatty acid concentrations were significantly lower and HDL cholesterol higher at troglitazone doses of 600 and 800 mg/day. LDL cholesterol increased at 400 and 600 mg doses only (from 4.3 and 3.9 mmol/1 at baseline to 4.8 and 4.5 mmol/1, respectively at 12 weeks, p < 0.05), but not at doses of 800 mg once daily or 400 mg twice daily. LDL/HDL ratio did not change during treatment. All doses were well tolerated; incidence of adverse events in troglitazone-treated patients was no higher than in those treated with placebo. However, a tendency to reduced neutrophil counts was observed in patients taking the highest doses of troglitazone. We conclude that troglitazone is effective and well-tolerated and shows potential as a new therapeutic agent for the treatment of NIDDM. [Diabetologia (1996) 39: 701-709] Kelncorfls Insulin resistance, thiazolidinedione, troglitazone, oral anti-diabetic agent, non-insulin-dependent diabetes mellitus.