Substantial evidence suggests that obesity is associated with non-insulin-dependent diabetes mellitus (NIDDM) and that weight reduction precedes improvements in glucose intolerance. However, achieving weight loss requires labour-intensive behavioural interventions. We evaluated the cost-utility of behavioural interventions in an experimental study of 76 NIDDM adults. In comparison with an education control group, adults randomly assigned to a diet-plus-exercise programme showed significant improvement in health status over an eighteen-month period. Using a general health policy model, we estimated that the programme produced 0.092 well-years for each participant. Programme costs were estimated using usual charges for physical examinations, blood tests, ECG evaluations, behaviour modification, and medical supervision. These costs totalled approximately US $1000 per participant. The cost-utility ratio was US $1000/0.092 = $10 870 per well-year. Using the general health policy model, the cost-utility of the behavioural intervention programmes was shown to be competitive with other widely advocated medical interventions.