Recent published statistics show that young, educated, male patrol officers with an average age of 5 years or less on the police force are most likely to be assaulted. However, analyses of data on assailant characteristics indicate that police officers are assaulted by individuals who exhibit a wide variation of physical and social attributes. Another approach to understanding attacks of police has involved analysis of the relationship between assault or murder rates of police and certain community characteristics. A discussion of research on circumstances does not reveal particular identifying characteristics of officers, offenders or communities likely to adequately explain violence against the police. Analysis of data pertaining to weapons used in attacks on police shows that almost all homicides of police involve firearms; and that assaults on police are characterized by the use of personal force. It is difficult, however, to determine if a difference between the weapons characteristic of assaults and homicides indicates a real difference in intent or motivation of the assailant. Although social-psychological studies suggest that police homicides may result from escalation of assault, the research required to make such an evaluation has not yet been conducted; nor have the social-psychological motives of persons who kill police been analyzed. The following methodological shortcomings of empirical studies relating to violence against police are described: (1) reporting procedures, (2) definitional inconsistencies, (3) bias of the respondents, (4) lack of representativeness of the findings, and (5) lack of comparison groups and baseline information. The implications for future research through correction of methodological deficiencies are discussed. It is concluded that further studies of assaults and homicides of police ought to incorporate some theoretical framework so that research implications may extend beyond a descriptive assessment of the dangers of policing. Footnotes are provided.