This is the first study to examine whether training before breakfast in the overnight‐fasted state is more effective in improving the health of patients with type 2 diabetes mellitus (T2DM) than after breakfast in the fed state. Thirty T2DM patients (60 ± 8 years, 33.7 ± 4.6 kg/m2) were randomly assigned to the F group (training in the overnight‐fasted state (n = 15)) and to the C group (training in the fed state (control group, n = 15)). All patients completed an 8‐week combined endurance/strength training program. Physical training significantly increased time to physical exhaustion during an endurance test (+10.4%), power output during strength tests (chest presses: +36.7% and seated rows: +37.8%), and fat‐free mass (+1.7 kg). Body fat mass (−1.9 kg), glycated hemoglobin (HbA1c) values (absolute change: −0.3%), serum insulin values (−2.5 microU/mL), the homeostatic model assessment for insulin resistance (HOMA‐IR) index (−1.1), and circulating triglyceride levels (−31 mg/dL) decreased significantly from pre‐ to post‐training. The training had no effect on body mass index, serum fasting glucose, total cholesterol, low‐density lipoprotein/high‐density lipoprotein ratio or interleukin (IL)‐6, IL‐10 and tumor necrosis factor (TNF)α levels. Analyses of variance revealed no time × group interaction for any variable (P > .05). The training was effective in improving the health of T2DM patients. However, the preliminary study's data do not provide any evidence that the nutritional state (overnight‐fasted or fed) in regular physical training plays a significant role for training‐induced adaptations in T2DM patients. Full trials (using other training protocols as well) should be conducted to gain further knowledge about the relevance of pre‐exercise breakfast ingestion.