OBJECTIVES: Patients with insulin-treated type 2 diabetes are more likely to be diagnosed with cancer than those on metformin. We looked for evidence of a relationship between insulin dose and cancer incidence in type 2 diabetes, and examined the effect of concomitant metformin upon this relationship. METHODS: Three treatment cohorts: metformin monotherapy, insulin plus metformin, and insulin only, were compared in a retrospective analysis of data from UK general practice. Relative insulin exposure was determined by the number of insulin prescriptions per year (<7, 7-10, 10-15, and >15). Cox proportional hazards models were used to adjust for differences in baseline characteristics and for differences in insulin exposure using a time-dependent analysis.  RESULTS: The adjusted hazard ratios (HRs) for a diagnosis of cancer in the four subgroups treated with insulin alone relative to metformin monotherapy, were: 1.24 (95% CI 0.90-1.70), 1.40 (1.05-1.87), 1.99 (1.52-2.61), and 2.75 (2.01-3.75) with increasing exposure to insulin. The corresponding HRs for insulin combined with metformin were 1.01 (0.72-1.40), 1.14 (0.85-1.52), 1.05 (0.76-1.46) and 1.40 (0.93-2.11). Introduced as a continuous variable, for those with insulin plus metformin the HR for prescribed insulin dose in IUs/kg/day was exactly 1.00 (0.76-1.29; p=0.9999). For insulin only, the HR for dose expressed in IUs/kg/day was 1.83 (1.50-2.24; p<0.0001). CONCLUSIONS: There was a dose-related association between insulin exposure and cancer risk in type 2 diabetes. The highest observed rates were in those with lower weight and higher levels of prescribed insulin. Co-administration of metformin was associated with a lower incidence of cancer at all levels of insulin exposure.