Ten insulin-dependent C-peptide-negative diabetic subjects, whose control had been optimized on twice-daily injection therapy, were treated for periods of 10 wk in a crossover study, with either a thrice-daily subcutaneous insulin injection regimen (Actrapid + Ultratard) or by continuous subcutaneous insulin infusion (CSII). On CSII insulin dose stabilized at 51 +/- 5 U/day, compared with 80 +/- 9 U/day (P = 0.004) on the thrice-daily injection regimen, having been 60 +/- 6 U/day on twice-daily therapy. After 10 wk glycosylated hemoglobin was 11.7 +/- 0.6% on injection therapy and 10.0 +/- 0.7% (P = 0.026) on CSII. Mean blood glucose concentration and urinary glucose excretion were lower at most points during the study on CSII than on injection therapy. Patients on pumps gained weight compared with the thrice-daily injection regimen (P = 0.023 at 10 wk) and the previous twice-daily regimen, despite the reduction in insulin dose. Considering individual patients, four markedly improved on CSII compared with the previous twice-daily regimen and five compared with Actrapid + Ultratard. No patient showed impaired control on CSII compared with either injection regimen. The benefits of portable insulin infusion pumps over injection therapy are thus clearly demonstrable under outpatient conditions even with equal and intensive medical attention.