BACKGROUND : Medication reconciliation is a powerful formal process to decrease medication errors, but it has proved to be complex and time consuming. OBJECTIVE : To describe the frequency and types of medication discrepancies (between previous treatment and medication order at admission), and to identify predictors of unintentional medication discrepancies (UMDs). METHODS : This interventional study was carried out in the cardiology department of a French teaching hospital. Medication reconciliation was conducted at admission to the cardiology department over 1 month in 2016 by trained pharmacists for: (1) determination of best possible medication history using multiple sources; (2) comparison with the patient's admission medication order and identification of discrepancies; and (3) classification of discrepancies (intentional/unintentional) with the physician. Associations between UMDs and various factors were examined. RESULTS : Overall, 100 patients were included (mean age 67.6±16.7 years; 56 men). The reconciliation process identified 544 drug discrepancies, 77 of which were UMDs; these occurred in 42 patients. The most common UMD type was omission (70.1%). Inability to speak French (P=0.007), low educational level (P=0.004), admission to a non-intensive care unit (P=0.019), two or more co-morbidities (P=0.001) and eight or more drugs on the admission order (P=0.004) were significantly associated with UMDs. Educational level remained significantly and independently associated with UMDs in a multivariable analysis after adjustment for factors that were statistically significant in the univariate analysis. CONCLUSIONS : This study highlights the high risk of medication discrepancies and the factors associated with UMDs. Our results allowed us to identify patients who should receive priority medication reconciliation in a cardiology department.