An improper balance of regulatory/effector T (Treg/Teff) cells is central to the development of autoimmune diseases, including type 1 diabetes. We previously showed that low-dose interleukin 2 (IL2) induced Treg cell expansion and activation and clinical improvement in patients with hepatitis-C-virus-induced vasculitis. We aimed to establish which low doses of IL2 would be safe and induce Treg cells in patients with type 1 diabetes, considering that: (1) type 1 diabetes might be linked to alteration of the IL2/IL2R activation pathway; (2) activation of pathogenic Teff cells by IL2 could exacerbate disease; and (3) the safety of low-dose IL2 is not known in type 1 diabetes. This was a single-centre phase 1/2 study. 24 adult patients (18-55 years) with established insulin-dependent type 1 diabetes and at least one diabetes-related autoantibody were enrolled and randomly assigned (in a 1:1:1:1 ratio, by computer-generated randomisation list, with block size four) to placebo or IL2 at 0.33 MIU/day, 1 MIU/day, or 3 MIU/day for a 5-day course and were followed up for 60 days. All investigators and participants were masked to assignment. The primary outcome was change in Treg cells, measured by flow cytometry, and expressed as a percentage of CD4+ T cells, from day 1 to day 60. This trial is registered with ClinicalTrials.gov, number NCT01353833. Six patients were assigned to each group between June 1, 2011, and Feb 3, 2012. IL2 was well tolerated at all doses, with no serious adverse events. However, there was a dose-response association for non-serious adverse events during the treatment phase (days 1-6); one patient in the placebo group, three patients in the 0.33 MIU group, five patients in the 1 MIU group, and six patients in the 3 MIU group had non-serious adverse events. The most common adverse events in the treatment phase were injection-site reaction (no patients with placebo vs three patients with 0.33 MIU and 1 MIU vs two patients with 3 MIU) and influenza-like syndrome (no patients with placebo vs one patient with 0.33 MIU and 1 MIU vs four patients with 3 MIU). After the treatment phase, adverse events did not differ between groups. IL2 did not induce deleterious changes in glucose-metabolism variables. IL2 induced a dose-dependent increase in the proportion of Treg cells, significant at all doses compared with placebo (placebo mean increase 0.5% [SD 0.4]; 0.33 MIU 2.8% [1.2], p=0.0039; 1 MIU 3.9% [1.8], p=0.0039; 3 MIU 4.8% [1.9] p=0.0039). We have defined a well-tolerated and immunologically effective dose range of IL2 for application to type 1 diabetes therapy and prevention, which could be relevant to other disorders in which a Treg cell increase would be desirable.