OBJECTIVE Adverse drug events represent the most common cause of preventable nonsurgical adverse events in medicine but may remain undetected. Our objective is to determine the proportion of drug-related visits emergency physicians attribute to medication-related problems. METHODS This prospective observational study enrolled adults presenting to a tertiary care emergency department (ED) during 12 weeks. Drug-related visits were defined as ED visits caused by adverse drug events. The definition of adverse drug event was varied to examine both narrow and broad adverse drug event classification systems. Clinical pharmacists evaluated all patients for drug-related visits, using standardized assessment algorithms, and then followed patients until hospital discharge. Interrater agreement for the clinical pharmacist diagnosis of drug-related visit was assessed. Emergency physicians, blinded to the clinical pharmacist opinion, were interviewed at the end of each shift to determine whether they attributed the visit to a medication-related problem. An independent committee reviewed and adjudicated all cases in which the emergency physicians' and clinical pharmacists' assessments were discordant, or either the emergency physician or clinical pharmacist was uncertain. The primary outcome was the proportion of drug-related visits attributed to a medication-related problem by emergency physicians. RESULTS Nine hundred forty-four patients were enrolled, of whom 44 patients received a diagnosis of the narrowest definition of an adverse drug event, an adverse drug reaction (4.7%; 95% confidence interval [CI] 3.5% to 6.2%). Twenty-seven of these were categorized as medication-related by emergency physicians (61.4%; 95% CI 46.5% to 74.3%), 10 were categorized as uncertain (22.7%; 95% CI 12.9% to 37.1%), and 7 categorized as a non-medication-related problem (15.9%; 95% CI 8.0% to 29.5%). Seventy-eight patients (8.3%; 95% CI 6.7% to 10.2%) received a diagnosis of an adverse drug event caused by an adverse drug reaction, a drug interaction, drug withdrawal, a medication error, or noncompliance. Emergency physicians attributed 49 of these to a medication-related problem (62.8%; 95% CI 51.7% to 72.7%), were uncertain about 15 (19.2%; 95% CI 12.0% to 29.4%), and attributed 14 to non-medication-related problems (17.9%; 95% CI 11.0% to 27.9%). Twenty-five of 29 (86.2%; 95% CI 69.3% to 94.4%) adverse drug events not considered medication related by emergency physicians were rated at least moderate in severity. CONCLUSIONS A significant proportion of drug-related visits are not deemed medication related by emergency physicians. Drug-related visits not attributed to medication-related problems by emergency physicians may be missed in ongoing outpatient adverse drug event surveillance programs intended to develop strategies to enhance drug safety. Further research is needed to determine what the effect may be of not attributing adverse drug events to medication-related problems.