This study assessed the safety profile and efficacy of a new combination therapy (insulin lispro plus sulfonylurea) in patients with type 2 diabetes mellitus experiencing secondary oral agent failure. A total of 423 patients were randomly assigned to 3 treatment groups: preprandial insulin lispro plus sulfonylurea (L + S), bedtime neutral protamine Hagedorn (NPH) insulin plus sulfonylurea (N + S), and preprandial insulin lispro plus bedtime NPH insulin (L + N). Mean decreases in glycosylated hemoglobin from baseline were 1.60%+/-1.27% for patients receiving L + S, 1.21%+/-1.21% for those receiving N + S, and 1.40%+/-1.46% for those receiving L + N (within treatment, P<0.001; for L + S vs. N + S, P = 0.003). Fasting blood glucose level was higher in patients receiving L + S (171+/-46.5 mg/dL) or L + N (166+/-52.5 mg/dL) than in those receiving N + S (144+/-48.2 mg/dL) (P<0.001, for both comparisons). Conversely, postprandial blood glucose level was lower in patients receiving L + S (165+/-41.6 mg/dL) or L + N (165+/-46.3 mg/dL) than in those receiving N + S (213+/-58.3 mg/dL) (P<0.001, for both comparisons). The overall rate of hypoglycemia (episodes per 30 days) was not statistically significant when the L + S, N + S, and L + N therapies were compared (0.99+/-1.74 vs. 0.87+/-2.31 vs. 1.16+/-2.38, respectively). The rate of nocturnal hypoglycemia was lowest in the L + S group (0.00+/-0.00 vs. 0.10+/-0.37 for the N + S group vs. 0.15+/-0.54 for the L + N group; P = 0.004). L + S, which has a safety profile equal to those of N + S and L + N, is an effective treatment for patients with type 2 diabetes who experience oral sulfonylurea agent failure. L + S offers an alternative to these established combination therapies in patients whose type 2 diabetes cannot be controlled with a sulfonylurea alone.