Most Danish hospitals have ward-based medication supply systems: frequently used drugs are stored in ward drug cupboards and dispensed by nursing staff. The objective of this study was to describe the frequency of dispensing errors and to identify risk factors.Cross-sectional study conducted on five hospital wards. Random samples of dispensed drugs were compared to the written drug orders.In total, 619 (94.4%) of the 656 samples were dispensed correctly (95% CI 92.3 to 95.9%). Among 2360 opportunities for error (2336 dispensed plus 24 prescribed but not dispensed doses), 42 errors were detected, error rate 1.8% (95% CI 1.3 to 2.4%). 55% of all errors were omissions. Logistic regression analysis revealed that dispensing errors were associated with the scheduled administration time, the ward and the number of dispensed doses. Having one rather than two nurses dispensing and administering drugs did not affect the error rate.For each scheduled time, approximately 19 of 20 patients' drugs are correctly dispensed. Although the clinical significance of errors was not rated, the advantage of having one rather than two nurses dispensing and administering is dubious. Reducing poly-pharmacy, however, might reduce the error rate and this strategy should be tested in a prospective study. As a direct consequence of this study, emphasis will be placed on analysing and learning procedures from wards with low error rates.