Development of hyperglycemia with subsequent ketoacidosis is one of the potential risks of a sudden cessation of insulin delivery during continuous insulin infusion therapy with insulin pumps in patients with IDDM. To evaluate differences in the development of ketoacidosis after a sudden pump stoppage between regular human insulin and insulin lispro, we performed an open label randomized crossover investigation with 7 patients (6 male/1 female, mean age (SD: 40.9 +/- 12.9 years). At 10 p.m., 4 hours after a light dinner with a preprandial injection of the corresponding insulin, the catheter was pulled out of the skin. During the observation period, blood glucose (every hour), pH-values and base excess values (every two hours) were measured until 7 a.m. One patient, in the insulin lispro treatment arm, discontinued because early interruption criteria were met after 7 hours. With insulin lispro, the metabolic changes developed 1.5 to 2 hours earlier than with regular human insulin (after 3 hours: difference in base excess (BE) mean +/- SD: regular human insulin: -0.41 +/- 1.04 mmol/l; insulin lispro: -1.69 +/- 0.83 mmol/l, p < 0.05; blood glucose: regular human insulin: 4.93 +/- 2.87 mmol/l, insulin lispro: 8.97 +/- 3.48, p < 0.05; pH values: regular human insulin: 7.38 +/- 0.02, insulin lispro: 7.36 +/- 0.02, n.s.). In general, metabolic deterioration tended to be more pronounced with insulin lispro than with regular human insulin (deltaBE after 7 h: regular human insulin: -2.39 +/- 1.30 mmol/l; insulin lispro: -3.27 +/- 2.43 mmol/l, n.s.). In conclusion, if patients want to be treated with insulin lispro in an insulin pump, they have to be well-educated about the pharmacokinetic properties of the insulin analogue and about the possibility that ketoacidotic deterioration after an interruption of the insulin delivery may occur earlier in comparison to regular human insulin. It is anyway recommendable to perform a pump stop test when starting CSII-treatment in patients with diabetes mellitus.