OBJECTIVE: To compare the plasma glucose (PG) response with a fixed mixture of 25% insulin lispro and 75% NPL (Mix25), prior to a meal and 3 h before exercise, to human insulin 30/70 (30/70) in patients with type 2 diabetes. RESEARCH DESIGN AND METHODS: Thirty-seven patients were treated in a randomized, open-label, 8-week, two-period crossover study. Mix25 was injected 5 min before breakfast and dinner throughout the study, as was 30/70 on inpatient test days and on outpatient dose titration days. Following the 4-week outpatient phase, patients were hospitalized, and exercised at a heart rate of 120 beats/min on a cycle ergometer two times for 30 min, separated by 30 min rest, starting 3 h after a 339 kcal breakfast. RESULTS: The 2-h postprandial PG was significantly lower with Mix25 ((mean +/- SEM) 10.5 +/- 0.4 mmol/l vs 11.6 +/- 0.4 mmol/l; p = 0.016). Maximum decrease in PG from onset of exercise to end of exercise was significantly less with Mix25 (-3.6 +/- 0.29 mmol/l vs -4.7 +/- 0.31 mmol/l; p = 0.001). The maximum decrease in PG over 6 h, after exercise onset, was significantly less with Mix25 (-4.3 +/- 0.4 mmol/l vs -5.9 +/- 0.4 mmol/l; p < 0.001). The frequency of hypoglycemia (blood glucose (BG) < 3 mmol/l or symptoms) during the inpatient test was not different between treatments. During the outpatient phase, the frequency of patient-recorded hypoglycemia was significantly lower with Mix25 (0.7 +/- 0.2 episodes/30 d vs 1.2 +/- 0.3 episodes/30 d; p = 0.042). CONCLUSIONS: Mix25 resulted in better postprandial PG control without an increase in exercise-induced hypoglycemia. The smaller decrease in PG during the postprandial phase after exercise may suggest a lower risk of exercise-induced hypoglycemia with Mix25 than with human insulin 30/70, especially for patients in tight glycemic control.