OBJECTIVE To compare the effects of insulin lispro (LP) and human regular insulin (HR) when given twice daily with NPH insulin on glycemic control (HbA1c), daily blood glucose profiles and rates of hypoglycemia in patients with type 2 diabetes mellitus after failure to respond to sulfonylurea drugs. METHODS A 5.5-month randomized, open-label, parallel study of 148 patients receiving either LP (n = 70) or HR (n = 78). Eight-point blood glucose profiles and HbA1c measurements were collected at baseline, 1.5, 3.5 and 5.5 months. RESULTS Two-hour post-breakfast and 2-hour post-supper blood glucose levels (means [and standard errors]) were significantly lower for LP than for HR at the end point (9.5 [0.4] mmol/L v. 10.9 [0.4] mmol/L and 8.4 [0.4] mmol/L v. 9.7 [0.4] mmol/L, respectively, p = 0.02 in both cases). HbA1c improved from 10.5% (0.2%) (LP) and 10.3% (0.2%) (HR) to 8.0% (0.1%). Hypoglycemia rates were similar during the day; however, there was an overnight trend to reduced rates with LP (0.08 [0.03] episodes/30 d v. 0.16 [0.04] episodes/30 d, p = 0.057). Quality-of life assessment showed significant improvement (p < 0.05) in the diabetes-related worry scale for LP subjects whereas HR subjects slightly worsened. CONCLUSIONS With traditional twice-daily insulin administration algorithms, LP improves 2-hour postprandial glucose levels, quality of life and overnight hypoglycemia rates while delivering an equivalent level of glycemic control (HbA1c) compared with HR to insulin-naïve patients with type 2 diabetes who require insulin.