We explored the epidemiology of hypoglycaemia in individuals with insulin-treated diabetes by testing the hypothesis that diabetes type and duration of insulin treatment influence the risk of hypoglycaemia. This was an observational study over 9-12 months in six UK secondary care diabetes centres. Altogether 383 patients were involved. Patients were divided into the following three treatment groups for type 2 diabetes: (1) sulfonylureas, (2) insulin for <2 years and (3) insulin for >5 years, and into two treatment groups for type 1 diabetes, namely <5 years disease duration and >15 years disease duration. Self-reported (mild and severe) and biochemical episodes (interstitial glucose <2.2 mmol/l using continuous glucose monitoring) were recorded. Mild hypoglycaemia in type 2 diabetic patients on insulin for <2 years was less frequent than in type 1 patients with <5 years disease duration (mean rate: 4 vs 36 episodes per subject-year, p < 0.001). In type 2 diabetic patients treated with sulfonylureas or insulin for <2 years, no differences were observed in the proportion experiencing severe hypoglycaemia (7 vs 7%, difference 0 [95% CI: -7 to 9%]), mild symptomatic (39 vs 51%, difference 12 [-3 to 25%]) or interstitial glucose <2.2 mol/l (22 vs 20%, difference 2 [-13 to 10%]). Severe hypoglycaemia rates were comparable in patients with type 2 diabetes on sulfonylureas or insulin < 2 years (0.1 and 0.2 episodes per subject-year) and far less frequent than in type 1 diabetes (<5 years group, 1.1; >15 years group, 3.2.episodes per subject-year). During early insulin use in type 2 diabetes, the frequency of hypoglycaemia is generally equivalent to that observed in patients treated with sulfonylureas and considerably lower than during the first 5 years of treatment in type 1 diabetes.