In order to evaluate the safety and effectiveness of insulin pump treatment and to establish criteria for its use, we retrospectively studied 45 patients from a referral university diabetes clinic who were treated with either intensive subcutaneous insulin injections or continuous subcutaneous insulin infusion. Hemoglobin A1C was 8.5 +/- 0.3% (SEM) before and 8.1 +/- 0.1% during continuous subcutaneous insulin infusion (p = NS), but rose to 10.0 +/- 0.2% at a 16-month post-study follow-up. The frequency of ketoacidosis was 0.17 events/year before and 0.20 during infusion pump use (p = NS) and declined to 0.10 at the follow-up (p = NS). Severe hypoglycemia was reduced from 2.73 events/year to 0.22 during continuous subcutaneous infusion treatment (p less than 0.001), and from 3.72-0.32 (p less than 0.001) in a subgroup of 23 patients who initiated insulin pump treatment because of frequent and severe hypoglycemic events during intensive insulin injection treatment. Insulin pump use, compared to intensive insulin injections in a non-research setting, (a) is equally effective for maintenance of near normal glycosylated hemoglobin levels, (b) need not result in increased ketoacidosis, and (c) is effective for reducing hypoglycemic events. Thus, insulin pump treatment can benefit larger and randomized studies are needed to confirm these results.