Six-month efficacy of benfluorex (Mediator) (150-450 mg/day) was assessed in a double-blind multicenter study vs. placebo or metformin hydrochloride (850-2550 mg/day). After a 2-month run-in period of strict dieting, 722 type 2 diabetic patients were randomized (1:2:2) to receive placebo (n=144), benfluorex (n=294) or metformin (n=284). After a 5-week dose-finding phase, the efficacy of benfluorex was compared with that of placebo (test for difference, main analysis) and metformin (non-inferiority test, secondary analysis) during a 6-month fixed-dose treatment. At entry after strict dieting, there was no difference between groups for HbA(1C) (placebo, 7.4%+/-1.5%; benfluorex, 7.7%+/-1.6%; metformin, 7.8%+/-1.6%) and fasting plasma glucose (FPG; placebo, 9.7+/-2.3 mmol/l; benfluorex, 10.0+/-2.0 mmol/l; metformin, 10.2+/-2.5 mmol/l). At the end of the dose-finding phase, mean doses were 2.71 tablets/day for placebo group, 2.65 tablets/day for benfluorex (397.5 mg/day) and 2.50 tablets/day for metformin (2125 mg/day). At the end of treatment, HbA(1C) level decreased by 0.60% ( p<0.001) in benfluorex patients while it increased by 0.50% ( p<0.001) with placebo (intent-to-treat analysis). The mean endpoint difference was -0.86% (SE, 0.17%; p<0.001). Mean endpoint difference in HbA(1C) between benfluorex and metformin was 0.28% (SE, 0.12%) [90% CI, 0.07 to 0.48] (non-inferiority test, p=0.037). Treatment with benfluorex was well tolerated; 39% of these patients reported one or more emergent adverse events (compared to 38% on placebo and 43% on metformin) and only two patients suffered a treatment-related, serious adverse event. This study demonstrates that benfluorex: (1) significantly reduces HbA1C and fasting plasma glucose when compared to placebo; (2) has a good safety profile; and (3) has relatively lower potency compared to metformin, although the non-inferiority test (equivalence limit for HbA(1C) of 0.5%) was significant.