Objective: To describe the outcome of 35 patients with type 2 diabetes prospectively followed for 6 years after the addition of a thiazolidinedione (TZD) to a failing regimen of a sulphonylurea and metformin - triple oral therapy. Methods: Study patients were assessed for the need for the addition of insulin to their regimen, and follow-up clinical and laboratory findings were analysed. Results: At a mean follow-up of 72±1.5 months (range 53-80), 18 (51%) of patients remained well controlled on triple oral therapy with a mean glycosylated haemoglobin (HbA1c) value of 6.9±0.2% (upper limit of normal 6.2%). In 17 other patients, triple oral therapy failed and the use of insulin was necessary after a mean duration of 38 (range 18-68) months. The mean HbA1c in these patients was 8.0±0.3%. The group that was maintained on triple oral therapy gained 15.2±1.9 lbs over the 6-year study which was significantly higher than the baseline weight. Alternatively, the group that failed and had insulin added to their therapy gained 20.2±4.5 lbs over the same period which was also significantly different from baseline but not from the triple oral therapy group. Although after 3 years a trend towards weight loss occurred in the triple oral therapy group, the insulin-added group continued to gain weight. Stimulated C-peptide levels increased significantly in the triple therapy group from 3.6±0.9 to 4.3±1.2 ng/ml and had not increased or decreased non-significantly from 3.7±0.8 to 3.2±0.6 ng/ml at the time of insulin initiation in the insulin-requiring group. Conclusion: When used late in the course of type 2 diabetes, TZDs result in improved and prolonged glycaemic control which persisted for a median time of 6 years. Weight gain with TZDs peaks and then plateaus (and even trends downwards) at 3 years, although the addition of insulin to a failing oral therapy regimen results in a further and continuing weight gain in spite of inferior glycaemic control. Continuing glycaemic control with triple oral therapy is dependent on preservation or augmentation of endogenous insulin production. © 2005 Blackwell Publishing Ltd.