Aims/hypothesis The aim of this study was to investigate the safety and efficacy of alum formulated glutamic acid decarboxylase GAD 65 (GAD-alum) treatment of children and adolescents with type 1 diabetes after 4 years of follow-up. Methods Seventy children and adolescents aged 10-18 years with recent onset type 1 diabetes participated in a phase II, double-blind, randomised placebo-controlled clinical trial. Patients identified as possible participants attended one of eight clinics in Sweden to receive information about the study and for an eligibility check, including a medical history. Participants were randomised to one of the two treatment groups and received either a subcutaneous injection of 20 μg of GAD-alum or placebo at baseline and 1 month later. The study was blinded to participants and investigators until month 30. The study was unblinded at 15 months to the sponsor and statistician in order to evaluate the data. At follow-up after 30 months there was a significant preservation of residual insulin secretion, as measured by C-peptide, in the group receiving GAD-alum compared with placebo. This was particularly evident in patients with <6 months disease duration at baseline. There were no treatment-related serious adverse events. We have now followed these patients for 4 years. Overall, 59 patients, 29 who had been treated with GAD-alum and 30 who had received placebo, gave their informed consent. Results One patient in each treatment group experienced an episode of keto-acidosis between months 30 and 48. There were no treatment-related adverse events. The primary efficacy endpoint was the change in fasting C-peptide concentration from baseline to 15 months after the prime injection for all participants per protocol set. In the GADalum group fasting C-peptide was 0.332±0.032 nmol/l at day 1 and 0.215 ± 0.031 nmol/l at month 15. The corresponding figures for the placebo group were 0.354± 0.039 and 0.184±0.033 nmol/l, respectively. The decline in fasting C-peptide levels between day 1 and month 1, was smaller in the GAD-alum group than the placebo group. The difference between the treatment groups was not statistically significant. In those patients who were treated within 6 months of diabetes diagnosis, fasting C-peptide had decreased significantly less in the GADalum group than in the placebo-treated group after 4 years.