The diabetogenic impact of ethanol remains as a focal point of basic and clinical investigations. In this study, Wistar rats were subjected to daily intragastric ethanol administration (10 ml/kg body weight injection with 0 (control), 10, 20 and 33 % (v/v) ethanol in the injections, respectively) for 19 weeks. At the end of the administration, we found that the fasting plasma glucose level of the 33 % (v/v) ethanol-loaded group was 18 % higher than the control. Insulin sensitivity was decreased in a dose-dependent manner in all the ethanol-loaded groups (r − 0·842, P < 0·001) during intraperitoneal insulin tolerance test. Necrotic/haemorrhagic injury was detected in the pancreas and islet β-cell mass was significantly reduced in the 33 % (v/v) ethanol-loaded rats by immunohistochemical and morphometric analysis. At the molecular level, we detected a dose-dependent attenuation of phosphatidylinositol 3-kinase activity (r − 0·956, P < 0·001) and GLUT-4 expression (GLUT-4 mRNA, r − 0·899, P < 0·001; GLUT-4 protein, r − 0·964, P < 0·001) in skeletal muscle. These results demonstrated that drinking is a conditional aetiological factor for diabetes and excessive ethanol intake is negatively associated with both insulin sensitivity and β-cell mass. The whole-body insulin resistance might result from the ethanol-induced insulin signalling defects in muscle.