The efficacy and safety of the preprandial injection of insulin lispro was compared with the oral administration of glibenclamide in patients with early type 2 diabetes. In this open-label, multicenter study, 143 patients with a glucagon-stimulated increase in C-peptide of at least 0.4 nmol/L were randomized to receive preprandial insulin lispro (LP) or glibenclamide (GB) for 26 weeks. Seventy-five patients received LP (51 male/24 female; age 40 to 70 years, duration of diabetes 4.4 +/- 2.9 years) and 68 patients received GB (39 male/29 female; age 39 to 70 years; duration of diabetes 4.3 +/- 3.4 years). After 12 weeks, mean 90 minute blood glucose excursions were 0.9 +/- 1.0 mmol/L for LP and 1.8 +/- 1.2 mmol/L for GB (p < 0.0001). After 24 weeks, mean blood glucose excursions were 1.0 +/- 1.1 mmol/L for LP and 1.7 +/- 1.2 mmol/L for GB (p = 0.002). Body weight decreased slightly from 87.2 +/- 2.3 to 86.5 +/- 12.2 kg in the LP group and increased from 84.1 +/- 13.7 to 84.4 +/- 13.3 kg in the GB group. LP versus GB induced changes from baseline to endpoint in fasting C-peptide (nmol/L), proinsulin and insulin levels (pmol/L) were - 0.2 +/- 0.4 versus - 0.1 +/- 0.6 (p = 0.04), - 11.2 +/- 26.0 versus - 1.1 +/- 17.3 (p = 0.03), and - 27.8 +/- 147.4 versus + 32.6 +/- 286.2 (not significant), respectively. HbA 1c at baseline was 7.5 +/- 1.0 % for LP and 7.7 +/- 1.2 % for GB and did not change significantly in either group during the investigation. No significant difference was observed between the groups with respect to hypoglycemic episodes. Treatment with LP improved postprandial blood glucose control more than GB without increasing body weight or hypoglycemic episodes. In addition, use of LP was associated with a decrease in fasting C-peptide and proinsulin levels, suggesting a potential down regulation of endogenous insulin production and improved proinsulin processing efficiency.