AIMS: To evaluate the acceptability and efficacy of an injection of insulin lispro, before an afternoon meal. METHODS: The subjects, 43 patients with Type 1 diabetes, 16 boys and 27 girls, aged 12.4 +/- 2.4 years, were randomly assigned to the treatment (n = 20) or the untreated control group (n = 23). The treatment was an injection of insulin lispro immediately before the afternoon meal. The control group had no injection. The treatment and the control group consumed identical types of meals for 2 months. The mean before-dinner blood glucose was measured during the last 2 weeks of the study. RESULTS: Injection of insulin lispro resulted in a significant reduction in the before-dinner blood glucose compared with the untreated control group (10.4 +/- 3.8 mmol/l vs. 14.7 +/- 3.9 mmol/l, respectively). The number of days on which the blood glucose was > 10 mmol/l was reduced by half in the insulin lispro group. The difference in HbA1c between baseline and endpoint differed slightly but significantly between the two groups, in boys. Treated patients ate the meal less frequently (11.4 +/- 3.0 times per 15 days) than the control patients (14.4 +/- 0.6 times per 15 days) and injected themselves with insulin 8.9 +/- 3.6 times per 15 days. The HbA1c increased significantly with the number of meals taken without injection. There was no statistically significant difference in the frequency of hypoglycaemia or changes in weight between the two groups. CONCLUSIONS: We conclude that an injection of insulin lispro before the afternoon meal can effectively lower the before-dinner blood glucose, and in boys also lowers the HbA1c. Patients were satisfied with the lower blood glucose before dinner, and did not find the insulin lispro injection difficult. However, compliance with the protocol procedures decreased during a subsequent 6-month period.