Background amp; objectives: To improve the interpretation of fatal intoxications by establishing fatal and non-fatal reference concentrations of metformin in postmortem femoral blood and to further evaluate risk factors associated with fatal metformin intoxication. Methods: All forensic autopsies in Sweden where metformin was detected in femoral blood 2011-2016 were identified in the National Board of Forensic Medicine databases (NFMD). The cases were classified as single substance intoxications, A (n = 22), multiple substance intoxications, B (N = 7) and postmortem controls, C (N = 13). The control group consisted of cases where metformin was detected, but the cause of death excluded the incapacitation by metformin or other substances. Strict inclusion criteria were used, and all postmortem cases were assessed by two independent reviewers. All other cases where the inclusion criteria of groups A-C where not met formed group O (N = 78). The forensic findings logged in the NFMD where linked to national registers whereby information on comorbidities, dispensed drugs and clinical data could be obtained. Results: The mean age was 66 +/- 10 years in the total study population and did not differ between the groups. The proportion of men was 64% in group A, 71% in B, 77% in C and 74% in group O. The median values of metformin in group A (48.5 mu g/g; range 13.0-210 mu g/g) and B (21.0 mu g/g; range 4.40-95.0 mu g/g) were significantly (p amp;lt; 0.001 and p = 0.015 respectively) higher than those of the control group C (2.30 mu g/g; range 0.70-21.0 mu g/g). The median concentration of metformin in group A and B was also significantly higher than in group O (4.60 mg/g; range 0.64-54.0 mu g/g) (p amp;lt; 0.001 and p = 0.040 respectively). The results suggest that intoxication with metformin as a cause of death should be considered when the postmortem femoral blood level exceeds about 10 mg/g, although higher levels may be seen in postmortem in cases without incapacitation. The metformin intoxication was confirmed to be intentional in 23% (n = 5) of the single intoxications. Underlying factors identified as important for the remaining fatal metformin intoxications included living alone, any contraindication for the use of metformin, known alcohol abuse and a history of stroke or cardiovascular disease. Conclusions: The reported post mortem femoral blood concentrations of metformin can hopefully contribute to a better interpretation of results in suspected poisonings and obscure cases. Living in a single household, history of cardiovascular disease and contraindications, predominantly alcohol abuse, were associated with fatal metformin intoxication. (C) 2019 Elsevier B.V. All rights reserved.