This trial was undertaken in order to evaluate the effects of endurance training on health cost in type 2 diabetes. 35 diabetic patients were randomly assigned to 2 groups: After 10 drop-outs, 15 followed a training program (8 sessions followed by training at home at the level of the ventilatory threshold VT) while 10 had only routine treatment. Both groups were followed over 1 year with evaluation at 30, 120, 240 and 365 days for health costs, blood pressure, and a standard maximal exercise test, glycemic and lipid equilibrium, 6-min walking test, and exercise (Voorips) and quality of life questionnaires. The effectiveness of training was confirmed in the trained group by an increase in the Voorips score (5.25±3.3 P<.001) and a lack of decrease in VO2max and Pmax while in the untrained group VO2max decreased slightly (−2.16±2.5 P=0.014). Thus trained subjects at the end of the study reached a higher percentage of the theoretical maximal power (P=0.041). The 6-min walking distance (472.2±98.9 vs 547.6±56.7 P=0.020) was also higher than in the control group. Blood pressure, lipid profile and glycemic control did not significantly improve during this period in either groups, due decreasing doses in treatments prescribed by their physicians. In the trained group there was no hospitalization, in contrast (P=0.047) with controls in whom there was 1.27±2.20 (ie, 0 to 5 days) of hospitalization. The total health cost over this period is lowered by 50% in the trained group (P=0.018). In conclusion, endurance training at the level of the VT significantly prevents the progressive decline in aerobic working capacity evidenced in untrained diabetics over this period of observation. It results in a marked reduction in health cost due to a decrease in treatment and fewer hospitalizations.