BACKGROUND: The strict limiting criteria for the use of metformin in diabetes mellitus stem largely from reports, in the 1970s, of mortality and lactic acidosis associated with phenformin. Data about metformin are less clear and are based mainly on case reports. The aim of this study was to evaluate the safety of continued use of metformin in patients with contraindications to this agent. PATIENTS: Some 393 patients with type 2 diabetes mellitus (serum creatinine 130-220 &mgr;mol/l) were studied. Among them were 266 patients with coronary heart disease (CHD), 94 with congestive heart failure (CHF), and 91 with chronic obstructive pulmonary disease (COPD), all of whom had been treated with metformin. The patients were randomized to either continue or to stop metformin and were then followed for 4 years. RESULTS: Analysis was by intention-to-treat. The patients who stopped taking metformin showed a rise in body mass index and in hemoglobin A1c significantly greater than those who continued the drug. There were no cases of lactic acidosis. Lactic acid values did not differ in the two groups and correlated only with serum creatinine and body mass index. Microvascular diabetic complications, cardiovascular events, and cardiovascular and total mortality were identical in the two groups. CONCLUSIONS: Diabetic patients who are treated with metformin and who tolerate the drug well may continue taking it, even when mild renal impairment develops, possibly up to serum creatinine levels of 220 &mgr;mol/l. There is also no apparent reason why patients with CHD, CHF, and COPD should discontinue metformin.