This study assessed the dose-dependent efficacy and safety of exenatide over 12 weeks in Japanese patients with type 2 diabetes suboptimally controlled despite therapeutic doses of sulfonylurea (SU), SU plus biguanide, or SU plus thiazolidinedione. Patients were randomly assigned to placebo (N = 40), 2.5 microg (N = 38), 5 microg (N = 37), or 10 microg (N = 38) exenatide administered subcutaneously twice daily (BID). Patients randomly assigned to 10 microg exenatide received 5 microg BID for the first 4 weeks, with the dose escalated to 10 microg BID for the final 8 weeks. Patients were 60.3 +/- 9.7 years old, with body mass index 25.3 +/- 4.3 kg/m(2) and hemoglobin A1c (HbA1c) 8.0 +/- 0.8%. Baseline-to-endpoint HbA1c changes (%) were +0.02 +/- 0.1 (placebo), -0.9 +/- 0.1 (2.5 microg), -1.2 +/- 0.1 (5 microg), and -1.4 +/- 0.1 (10 microg) (all p < 0.001 vs. placebo). Of patients with baseline HbA1c -7%, 5.1% (placebo), 50.0% (2.5 microg), 71.4% (5 microg), and 79.4% (10 microg) achieved HbA1c <7% at endpoint (p < 0.001, trend test). Baseline-to-endpoint fasting plasma glucose changes (mg/dL) were +6.0 +/- 4.8 (placebo), -18.6 +/- 5.7 (2.5 microg), -25.0 +/- 7.0 (5 microg), and -28.9 +/- 5.9 (10 microg) (all p < or = 0.001 vs. placebo). Treatment-emergent adverse events were mostly mild; dose-dependent increases in incidence were observed for hypoglycemia, nausea, anorexia, decreased appetite, and diarrhea (all p < or = 0.044, trend test). Over 12 weeks, exenatide dose-dependently improved glycemic control in Japanese patients with type 2 diabetes.