Aims: The choice of insulin at initiation in type 2 diabetes remains controversial. The aim of this study was to assess the occurrence of self-reported severe hypoglycaemia associated with premixed insulin analogues in routine clinical care. Methods: A 12-month, prospective, observational, multicentre study in patients starting a commonly prescribed premixed insulin analogue (either insulin lispro 25/75 or biphasic insulin aspart 30/70, twice daily) after suboptimal glycaemic control on oral antidiabetic agents. Treatment decisions were made solely in the course of usual practice. Results: Study follow-up was completed by 991 (85.5%) of the 1150 patients enrolled. At baseline, mean (SD) age was 57.9 (10.1) years; mean diabetes duration was 9.2 (5.9) years; mean haemoglobin A1c (HbA1c) was 9.9 (1.8) % and the rate of severe hypoglycaemia was 0.03 episode/patient-year. At 12 months, the rate of severe hypoglycaemia was 0.04 episode/patient-year (95% CI 0.023, 0.055 episode/patient-year) and mean insulin dose was 41.5 (19.4) units. Changes from baseline to 12 months for mean fasting plasma glucose and HbA1c were -5.1 mmol/l and -2.5%, respectively. Conclusions: After initiation of premixed insulin analogues in patients with type 2 diabetes in real-world settings, the incidence of severe hypoglycaemia was lower than expected from previously reported studies.