It is widely believed among police officers that domestic incidents are among the most dangerous incidents to which they respond. However, most research in this area suffers from the “denominator problem,” where prior studies have focused on incidents resulting in harm to police officers and failed to account for incidents not resulting in harm. Such methodologies can produce drastically misleading results. This paper uses data from the 2016 National Incident Based Reporting System (NIBRS) to overcome the denominator problem. We examine the probability of (1) an officer being assaulted and (2) an officer being injured or killed when responding to a domestic incident compared to a non-domestic incident while controlling for other potentially important variables. Results indicate that officers are significantly more likely to be assaulted or injured when responding to non-domestic incidents. Implications for law enforcement training, victim legitimacy, and future research are discussed.