AimsIn this study, we assessed the effects of a 12‐week combined aerobic–resistance training and subsequent detraining on Beck Depression Inventory (BDI) score and mediating role of BDNF and also investigated whether exercise‐induced alterations are maintained following a short period of detraining in women with type 2 diabetes (T2D).Materials and MethodsThirty‐four women with T2D were randomly assigned to experimental or control group (age: 60.6 ± 6.3, body mass index (BMI): 30.2 ± 1.3 kg/m2, HbA1c: 8.09 ± 0.73%). The exercise training comprised of combined aerobic–resistance programme (50%–70% heart rate reserve for aerobic exercise, and 50%–70% 1 repetition maximum for resistance exercise, respectively) performed three sessions per week over 12 weeks. The intervention period was followed by an 8‐week detraining period. Data were collected at baseline and also following exercise intervention and detraining. Data were analysed by linear mixed model at p < 0.05.ResultsAfter 12 weeks of combined exercise training and 8 weeks of detraining, there was a significant difference in BDNF (0.08; 95% confidence interval [CI] = 0.07–0.10; p = 0.001), fasting blood glucose (FBG) (−45.41; CI = −50.83, ‐39.98; p = 0.001), insulin (−6.47; CI = −7.04, −5.9; p = 0.001), HOMA‐IR (−3.76; CI = −4.07, −3.45; p = 0.001) and BDI score (−17.17; CI = −20.29, −14.05; p = 0.001) between the experimental and control group. Multiple mediation analysis indicated that BDNF seems to have a mediating role in exercise‐induced improvement of depression (p = 0.04). After the detraining period, BDI score remained unchanged and it showed a significant increase compared to before the start of training (p = 0.001).ConclusionsIt may be concluded that exercise training improves depression that is likely to be explained by increased BDNF concentration in TD2. In spite of decreased BDNF concentration following an 8‐week detraining, depression score was maintained.