AIMS: Insulin analogues are widely used but few data exist comparing different analogue regimens. We compared two such regimens in type 2 diabetes mellitus (T2DM) uncontrolled by oral antidiabetic agents (OADs) with or without basal insulin. METHODS: In a 26-week multinational, multicentre, randomized treat-to-target trial, OADs were discontinued and subjects randomized to analogue basal-bolus therapy (insulin detemir once daily and insulin aspart mealtimes) or biphasic insulin aspart 30 (30% rapid-acting insulin aspart), twice daily. Insulin was titrated to targets for fasting, predinner and postprandial plasma glucose (PG), as appropriate. RESULTS: Of 719 subjects, 92% completed the study; 58% achieved haemoglobin fraction A(1c) (HbA(1c)) < or =7.0%, with reductions of 1.56% (to 6.96%) with basal-bolus therapy and 1.23% (to 7.17%) with biphasic insulin aspart. Reduction with basal-bolus therapy was superior in the overall population by 0.23% (p = 0.0052), with no difference between regimens in insulin-naive patients. Major hypoglycaemia occurred in five basal-bolus patients (0.9%) and in no patients with biphasic insulin aspart. Incidence of minor hypoglycaemia was similar in both groups. All insulin doses increased during titration, with increase in lunchtime insulin aspart dose and equal distribution of breakfast and dinner biphasic insulin aspart doses. Insulin detemir remained once daily in 87% of patients. CONCLUSIONS: Modern insulin analogue regimens, adjusted to PG targets, enable a majority of people with T2DM to reach HbA(1c)< or =7.0% after failure of OADs and OAD-basal insulin therapy. Insulin-treated patients may benefit more from transfer to analogue basal-bolus therapy, while insulin-naive individuals benefit equally well from the more convenient biphasic analogue regimen.