This study used Baltimore County Police Department (BCPD) assault data originally collected by Uchida and Brooks (1990). Information on 1,550 nonlethal police assaults that occurred from January 1984 through December 1986 was obtained from BCPD official records. The unit of analysis was the individual assault, defined as "any overt physical act that the officer perceives or has reason to believe was intended to cause him harm." Categories of variables measured were officer and assailant attributes, situational characteristics, and environmental characteristics. Logistic regression was used to estimate the likelihood of officer injury. A number of policy implications were derived from the findings. First, the data show that the majority of assault incidents involved unarmed attacks against the police and that such attacks were more likely to result in officer injury than armed attacks; moreover, most officers responded to assailants with physical force. Batons, firearms, or other weapons were rarely used. Thus, greater officer proficiency in unarmed defensive tactics may help reduce the number of police assault-related injuries. Second, since fewer years of service were associated with increases in risk of injury, police departments might consider providing additional inservice training for patrol officers with less than 5 or 6 years of experience. The findings also show that very short and very tall officers were more likely to be injured when assaulted than officers of average height. This suggests that defensive- tactics instructors may need to tailor additional strategies to increase these officers' effectiveness during use-of-force encounters. Fourth, substantial numbers of assaulted officers reported being victims multiple times within the study period, and actions taken by assailants prior to assaulting officers show that the risk of injury was, to some degree, associated with suspect motivation. Other findings were that officers assaulted by nonwhite suspects were more likely to be injured than officers assaulted by white suspects; officers who were college graduates were less likely to be injured than officers without a degree. Domestic disturbances were found to be no more dangerous than traffic stops, legal interventions, or "other" calls for service and were less dangerous than general disturbances. Policy implications are drawn for each of these findings. 2 figures, 3 tables, 51 notes, 79 references, and appended supplementary data