To assess how an isolated change in the pattern of care influences outcome of care and hospital use, a randomised prospective 2-year study was done in which 31 of 61 consecutive children with newly diagnosed insulin-dependent diabetes mellitus (IDDM) were admitted to hospital at disease onset for about a week and compared with the other 30 children who were admitted for about 4 weeks. Insulin treatment and education about diabetes were similar in the two groups. Duration of initial stay in hospital had no effect on metabolic control during the 2 years but time since diagnosis was significant with respect to effect on haemoglobin A1 (p = 0.001), haemoglobin A1c (p = 0.004), and insulin dose (p less than 0.001). At 2 years, 45% of the children in the short-term group and 29% in the long-term group were C-peptide positive (p = NS); C-peptide positivity correlated with age. A change in the pattern of care of children with IDDM, led to a pronounced decrease in hospital use by this patient group. Irrespective of the length of initial stay in hospital, equally good metabolic control was obtained in both groups for 2 years.