OBJECTIVE: To compare the efficacy of the short-acting insulin analog lispro (LP) with that of regular insulin in IDDM patients treated with an external pump. RESEARCH DESIGN AND METHODS: Thirty-nine IDDM patients (age, 39.4 ± 1.5 years; sex ratio, 22M/17W; BMI, 24.4 ± 0.4 kg/m2; diabetes duration, 22.5 ± 1.6 years) who were treated by external pump for 5.1 ± 0.5 years were involved in an open-label, randomized, crossover multicenter study comparing two periods of 3 months of continuous subcutaneous insulin infusion with LP or with Actrapid HM, U-100 (ACT). Boluses were given 0–5 min (LP) or 20–30 min (ACT) before meals. Blood glucose (BG) was monitored before and after the three meals every day. RESULTS: The decrease in HbA1c was more pronounced with LP than with ACT (−0.62 ± 0.13 vs. −0.09 ± 0.15%, P = 0.01). BG levels were lower with LP (7.93 ± 0.15 vs. 8.61 ± 0.18 mmol/l, P < 0.0001), particularly postprandial BG levels (8.26 ± 0.19 vs. 9.90 ± 0.20 mmol/l, P < 0.0001). Standard deviations of all the BG values (3.44 ± 0.10 vs. 3.80 ± 0.10 mmol/l, P = 0.0001) and of postprandial BG values (3.58 ± 0.10 vs. 3.84 ± 0.10 mmol/l. P < 0.02) were lower with LP. The rate of hypoglycemic events defined by BG < 3.0 mmol/l did not significantly differ between LP and ACT (7.03 ± 0.94 vs. 7.94 ± 0.88 per month, respectively), but the rate of occurrences of very low BG, defined as BG < 2.0 mmol/l, were significantly reduced with LP (0.05 ± 0.05 vs. 0.47 ± 0.19 per month, P < 0.05). At the end of the study, all but two (95%) of the patients chose LP for the extension phase. CONCLUSIONS: When used in external pumps, LP provides better glycemic control and stability than regular insulin and does not increase the frequency of hypoglycemic episodes.