Background: Errors may be more common and more likely to be harmful with opioids than with other medications, but little research has been conducted on these errors. Methods: The authors retrospectively analyzed MEDMARX®, an anonymous national medication error reporting database, and quantitatively described harmful opioid errors on inpatient units that did not involve devices such as patient-controlled analgesia. The authors compared patterns among opioids and qualitatively analyzed error descriptions to help explain the quantitative results. Results: The authors included 644 harmful errors from 222 facilities. Eighty-three percent caused only temporary harm; 60 percent were administration errors and 21 percent prescribing errors; and 23 percent caused underdosing and 52 percent overdosing. Morphine and hydromorphone had a significantly higher proportion of improper dose errors than other opioids (40 percent and 41 percent compared with 22 percent with meperidine). Hydromorphone errors were significantly more likely to be overdoses (78 percent vs 47 percent with other opioids). Omission errors were significantly more common with fentanyl patches (36 percent compared with 12 percent for other opioids). Wrong route errors were significantly more common with meperidine (given intravenously when prescribed as intramuscular, 34 percent vs 3 percent for morphine). Oxycodone errors were significantly more likely to be wrong drug errors (24 percent vs 11 percent for other opioids), often because of confusion between immediate- and sustained-release formulations. Conclusions: Reported opioid errors are usually associated with administration and prescribing and frequently cause uncontrolled pain as well as overdoses. These patterns of errors should be considered when using opioids and incorporated into pain guidelines, education, and quality improvement programs.