AIM: The role of metformin in lactic acidosis is regularly questioned. Arguments against a causal role for metformin in lactic acidosis occurrence are the lack of correlation between plasma metformin and lactate levels, as well as between metformin plasma levels and mortality. We aim to analyse these correlations in a large series of lactic acidosis cases recorded in the French nationwide pharmacovigilance database. METHODS: All cases of lactic acidosis spontaneously reported between 1985 and October 2013 associated with metformin exposure were extracted from the pharmacovigilance database. We assessed the statistical correlations between prescribed daily doses of metformin, plasma concentrations of metformin and lactate, pH and plasma creatinine, as well as the relationship between mortality and these variables. RESULTS: Seven hundred and twenty-seven cases of lactic acidosis were reported during the period. Metformin plasma concentration was documented for 260 patients, lactate plasma concentration for 556 patients, pH for 502 patients, creatinine for 397 patients and the vital outcome for 713 patients. Metformin plasma concentration, lactate concentration, pH and plasma creatinine were all correlated (P < 0.001). There were significant differences between surviving and deceased patients in terms of metformin plasma levels (25.2 vs. 37.4 mg/l, P = 0.002) and lactate concentrations (10.8 vs. 16.3 mmol/l, P < 0.001). Thirty per cent of patients died when metformin concentration was > 5 mg/l compared with 11% for patients with concentration < 5 mg/l (P = 0.003). CONCLUSIONS: Our data suggest that metformin accumulation contributes to the pathogenesis and prognosis of lactic acidosis.