OBJECTIVE: To compare safety and efficacy of biphasic insulin aspart 70/30 (BIAsp 30) with exenatide in subjects with type 2 diabetes mellitus (T2DM) not achieving glycemic targets with metformin and sulfonylurea in a randomized, open-label, 24-week trial. RESEARCH DESIGN AND METHODS: Subjects (N = 372, T2DM > 6 months, age > or = 18 and < or = 80 years, HbA1c > or = 8%, insulin naive not achieving glycaemic targets, receiving metformin and sulfonylurea) were randomized 1: 1: 1 to receive either BIAsp 30 QD (12 U before supper); BIAsp 30 BID (12 U divided equally between pre-breakfast and pre-supper); or exenatide (5 microg BID for 4 weeks and 10 microg BID thereafter). Efficacy (HbA1c, fasting plasma glucose [FPG]) and safety (adverse events and hypoglycemic episodes) were assessed. RESULTS: Glycemic control achieved with both BIAsp 30 BID and BIAsp 30 QD was superior to that with exenatide (BIAsp 30 BID-exenatide: HbA1c difference -0.91% [95% CI: -1.23 to -0.59%] and BIAsp 30 QD-exenatide: difference: -0.67% [95% CI: -0.99 to -0.34%]). At the end of the study, more subjects achieved HbA1c < 7% and < or = 6.5% in the BIAsp 30 BID group than in the exenatide group (HbA1c < 7%: 37% vs. 20%, p = 0.0060; HbA1c < or = 6.5%: 25% vs. 8%, p = 0.0004, respectively). Combined hypoglycemic episodes (major, minor, symptoms only) were reported by 56%, 61%, and 29% of the subjects in the BIAsp 30 QD, BIAsp 30 BID, and exenatide groups, respectively. Weight gain was observed in the BIAsp 30 group (BIAsp 30 QD: 2.85 kg, BIAsp 30 BID: 4.08 kg) and weight loss was observed in the exenatide group (-1.96 kg). Nausea or vomiting was responsible for discontinuation of seven subjects in the exenatide group and one subject in the BIAsp 30 BID group. CONCLUSIONS: Significantly more T2DM patients (poorly controlled with combination metformin/sulfonylurea) achieved glycemic goals when treated with BIAsp 30 than with exenatide. The high baseline HbA1c values (approximately 10.2%) and the long duration of diabetes (approximately 9 years) suggests that some subjects may have been in an advanced stage of their diabetes and may not have had sufficient beta-cell function for a GLP-1 mimetic to be effective. The insulin-treated groups had more minor hypoglycemic events and weight gain but less gastrointestinal side-effects. In summary, BIAsp 30 was more efficacious in helping patients with high baseline HbA1c achieve glycemic goals. CLINICAL TRIAL REGISTRATION: www.clinicaltrials.gov, NCT00097877.