The current study investigated the effects of change blindness and crime severity on eyewitness identification accuracy. This research, involving 717 subjects, examined change blindness during a simulated criminal act and its effects on subjects' accuracy for identifying the perpetrator in a photospread. Subjects who viewed videos designed to induce change blindness were more likely to falsely identify the innocent actor relative to those who viewed control videos. Crime severity did not influence detection of change; however, it did have an effect on eyewitness accuracy. Subjects who viewed a more severe crime ($500 theft) made fewer errors in perpetrator identification than those who viewed a less severe crime ($5 theft). This research has theoretical implications for our understanding of change blindness and practical implications for the real-world problem of faulty eyewitness testimony.