This research project was the outgrowth of a study conducted by one of the researchers on behalf of the plaintiffs in a lawsuit against the police department of a midsized, midwestern city. The data set consisted of 376 assault cases from the official records of the police department. It covered all cases designated by the department as an "assault" for a period of 10 months during 1986 to 1987, after a domestic violence reform statute was passed to emphasize police aggressive intervention in domestic assault cases. The study conducted a content analysis of 1,000 consecutively numbered incidents to obtain a general understanding of how domestic assault reports were written compared to other calls. The study found differential treatment of domestic violence and stranger assault incidents as measured by the likelihood of arrest. Despite the relatively low percentage of arrest (26 percent), it was clear that officers made fewer arrests for cases that involved domestic assaults than for stranger assault cases. This distinction was magnified when several relevant factors -- offender presence at the scene and victim preference -- were controlled. Of equal importance, victims' preferences for arrest were ignored in 75 percent of the domestic cases, compared with more than 40 percent of stranger assault cases. Although these findings suggest that police disparately respond to the problems of the largely female victims of domestic violence, the researchers are unwilling to state that this is symptomatic of overall police behavior. 4 tables, 5 notes, and 57 references