Objective:To characterize medication incidents occurred in an outpatient emergency service. Method: Descriptive, documental, retrospective and quantitative research. The International Classification for Patient Safety was the theoretical reference for the construction of the instrument used to collect and analyze the data from 119 notification and investigation forms of incidents occurred in 2014 in a teaching hospital. Data were collected twice, compared, corrected and transcribed to an Excel worksheet. The SPSS 19.0 Software and the non-parametric Mann-Whitney test were used in the analysis; p<0.05 indicated statistical significance. Results: A total of 142 incidents were analyzed, most of them involving the nursing team; 93.7% were avoidable; one-third involved high-alert medications; the majority involved parenteral administration. Harm was rare but proportional to the time elapsed for error detection. Management failures prevailed, especially omission. Conclusion: Most of the incidents analyzed were characterized as potentially harmful and avoidable, with emphasis on personnel factors as contributors.