BACKGROUND: Medication errors (ME) are major public health issues in hospitals because of their consequences on patients' morbi-mortality. This study aims to evaluate the prevalence of ME at admission and discharge of hospitalization in diabetic and non-diabetic patients, and determine their potential clinical impact. METHOD: This prospective observational study was conducted at the Endocrinology-Diabetology-Nutrition Department. All adult patients admitted were eligible. A total of 904 patients were included, of which 671 (74.2%) with diabetes mellitus. Clinical pharmacists conducted medication reconciliation: they collected the Best Possible Medication History and then compared it with admission and discharge prescriptions to identify medication discrepancies. ME were defined as unintended medication discrepancies if corrected by the physician. RESULTS: Clinical pharmacists allowed correcting ME in 176/904 (19.5%) patients at admission and in 86/865 (9.9%) patients at discharge. More than half of ME were omissions. Diabetic patients were more affected by ME than non-diabetic patients, both at admission (22.1% vs 12.0%, p<0.001) and at discharge (11.4% vs 5.7%, p=0.01). The diabetic group also had more potentially severe and very severe ME. Diabetic patients had on average twice more medications than non-diabetic patients (8.7±4.5 vs 4.4±3.4, p<0.001). The polypharmacy associated with diabetes, but not diabetes mellitus itself, was identified as a risk factor of ME. CONCLUSIONS: The intervention of clinical pharmacists allowed correcting 378 ME in 25.8% of the cohort before they caused harm. Clinicians, pharmacists and other health care providers should therefore work together to improve patients' safety, in particular in high-risk patients such as diabetic patients.