The short-acting insulin analogue lispro ([LYS(B28), PRO(B29)] is absorbed from the subcutis more rapidly than soluble insulin (S). To compare the clinical effectiveness of lispro vs S, 11 Type 1 patients using continuous subcutaneous insulin infusion (CSII) therapy (6 F, 5 M, age 30 +/- 2.5 years, diabetes duration 14 +/- 1.0 years, BMI 24.0 +/- 0.8 kg m(-2), HbA1c 6.5 +/- 0.2%) were studied in an open, randomized, crossover study for 6 months (3 months lispro and 3 months S or vice versa). During lispro treatment mean fasting and 2 h postprandial blood glucose were lower compared to the S phase (fasting 6.5 +/- 0.4 vs 7.5 +/- 0.4 mmol l(-1) (NS), postprandial 6.8 +/- 0.3 vs 8.3 +/- 0.3 mmol l(-1), p = 0.03). In patients treated first with lispro HbA1c levels improved from 6.3 +/- 0.2% to 5.7 +/- 0.3%; On reversion to S HbA1c increased to 6.2 +/- 0.2%. In the group treated first with S, HbA1c fell (6.7 +/- 0.4% vs 6.5 +/- 0.3%) and then improved further to 6.3 +/- 0.3% with lispro. None of these changes were significant. There was no significant difference with respect to hypoglycaemic or other adverse events. It can be concluded that lispro in CSII therapy is safe and may improve postprandial glucose excursions.