OBJECTIVE: To systematically review empirical evidence on the prevalence and nature of medication administration errors (MAEs) in health care settings. DATA SOURCES: Ten electronic databases (MEDLINE, EMBASE, International Pharmaceutical Abstracts, Scopus, Applied Social Sciences Index and Abstracts, PsycINFO, Cochrane Reviews and Trials, British Nursing Index, Cumulative Index to Nursing and Allied Health Literature, and Health Management Information Consortium) were searched (1985-May 2012). STUDY SELECTION AND DATA EXTRACTION: English-language publications reporting MAE data using the direct observation method were included, providing an error rate could be determined. Reference lists of all included articles were screened for additional studies. DATA SYNTHESIS: In all, 91 unique studies were included. The median error rate (interquartile range) was 19.6% (8.6-28.3%) of total opportunities for error including wrong-time errors and 8.0% (5.1-10.9%) without timing errors, when each dose could be considered only correct or incorrect. The median rate of error when more than 1 error could be counted per dose was 25.6% (20.8-41.7%) and 20.7% (9.7- 30.3%), excluding wrong-time errors. A higher median MAE rate was observed for the intravenous route (53.3% excluding timing errors (IQR 26.6-57.9%)) compared to when all administration routes were studied (20.1%; 9.0-24.6%), where each dose could accumulate more than one error. Studies consistently reported wrong time, omission, and wrong dosage among the 3 most common MAE subtypes. Common medication groups associated with MAEs were those affecting nutrition and blood, gastrointestinal system, cardiovascular system, central nervous system, and antiinfectives. Medication administration error rates varied greatly as a product of differing medication error definitions, data collection methods, and settings of included studies. Although MAEs remained a common occurrence in health care settings throughout the time covered by this review, potential targets for intervention to minimize MAEs were identified. CONCLUSIONS: Future research should attend to the wide methodological inconsistencies between studies to gain a greater measure of comparability to help guide any forthcoming interventions. © 1967-2013 Harvey Whitney Books Co. All rights reserved.