Weight change and glycaemic control in 132 diabetic patients previously treated for 1 year by four different methods aimed at weight loss were reanalysed 4 years after the outset. The four treatment groups comprised clinic visits, home visits, behavioural group therapy, and dexfenfluramine given for the initial 3 months followed by clinic visits. When analysed on an intention to treat basis only the dexfenfluramine group maintained a significant weight loss at year 4 (mean -2.46 kg) with 43% of patients losing 3 kg or more; HBA1c was not decreased. The other treatment groups showed overall weight regain from year 1 to year 4. Within the home visit group however, the number of patients losing at least 3 kg doubled between years 1 (21%) and 4 (38%). When analysed on a completion basis, weight loss in the dexfenfluramine group was significant in females but not in males at year 4. During the 4 years of observation a cohort of 54 patients reflecting our routine clinic practice gained on average 0.35 kg. Although now withdrawn, the use of an appetite suppressant dexfenfluramine for just 3 months would appear to have a long-term advantage on weight loss in this diabetic population, although the weight lost was not associated with improvement in glycaemic control.