Aim/hypothesis In people with type 2 diabetes, exercise improves glucose control (as reflected in HbA 1c ) and physical fitness, but it is not clear to what extent these exercise-induced improvements are correlated with one another. We hypothesised that reductions in HbA 1c would be related: (1) to increases in aerobic fitness and strength respectively in patients performing aerobic training or resistance training; and (2) to changes in strength and aerobic fitness in patients performing aerobic and resistance training. Methods We randomly allocated 251 type 2 diabetes patients to aerobic, resistance, or aerobic plus resistance training, or to a sedentary control group. Peak oxygen consumption (V : O 2peak ), workload, treadmill time and ventilatory threshold measurements from maximal treadmill exercise testing were measured at baseline and 6 months.Muscular strength was measured as the maximum weight that could be lifted eight times on the leg press, bench press and seated row exercises. Results With aerobic training, significant associations were found between changes in both V : O 2peak (p=0.040) and workload (p=0.022), and changes in HbA 1c. With combined training, improvements in V : O 2peak (p = 0.008), workload (p=0.034) and ventilatory threshold (p=0.003) were significantly associated with changes in HbA 1c. Increases in strength on the seated row (p=0.006) and in mid-thigh muscle cross-sectional area (p=0.030) were significantly associated with changes in HbA 1c after resistance exercise, whereas the association between increases in muscle cross-sectional area and HbA 1c in participants doing aerobic plus resistance exercise (p= 0.059) was of borderline significance. Conclusions/interpretation There appears to be a link between changes in fitness and HbA 1c . The improvements in cardiorespiratory fitness with aerobic training may be a better predictor of changes in HbA 1c than improvements in strength.