Aims/hypothesis: We evaluated and compared the effects on glycaemic control of two different exercise protocols in elderly men with type 2 diabetes mellitus. Methods: Eighteen patients with type 2 diabetes mellitus carried out home-based bicycle training for 5 weeks. Patients were randomly assigned to one of two training programmes at 60% of maximal oxygen uptake: three 10 min sessions per day (3 x 10) or one 30 min session per day (1 x 30). Plasma insulin, C-peptide and glucose concentrations were measured during a 3 h oral glucose tolerance test (OGTT). Insulin sensitivity index (ISI _composite), pre-hepatic insulin secretion rates (ISR) and change in insulin secretion per unit change in glucose concentrations (B_total) were calculated. Results: Cardiorespiratory fitness increased in response to training in both groups. In group 3 x 10 (n = 9) fasting plasma glucose (p = 0.01), 120 min glucose OGTT (p = 0.04) and plasma glucose concentration areas under the curve at 120 min (p < 0.04) and 180 min (p = 0.07) decreased. These parameters remained unchanged in group 1 x 30 (n = 9). No significant changes were found in ISI_composite, ISR and B_total in either of the exercise groups. In a matched time-control group (n = 10), glycaemic control did not change. Conclusions/interpretation: Moderate to high-intensity training performed at 3 x 10 min/day is preferable to 1 x 30 min/day with regard to effects on glycaemic control. This is in spite of the fact that cardiorespiratory fitness increased similarly in both exercise groups. A possible explanation is that the energy expenditure associated with multiple short daily sessions may be greater than that in a single daily session.