The research sought to determine whether assailants sometimes kill rather than merely injure their victims to avoid either retaliation or criminal prosecution. The study tested the hypotheses that, for these tactical reasons, offenders will be more likely to kill when they have no accomplices, when their victims are male or black, and when the victim can identify them. The data came from nonfatal incidents reported in the National Crime Victimization Survey covering the period 1987-91. The homicide statistics came from the FBI's 1989 Supplementary Homicide Reports. The data were analyzed using multiple logistic regression techniques. Results were largely consistent with the theoretical expectations. Lethal intent was more likely to be a factor in pure assault than in robbery or rape. The victim's compliance was evidently more important to robbery and rape offenders; harm to the victim was often incidental in these cases. Lethal outcomes were also much more likely when the offenders used guns. The use of knives was also associated with lethal outcomes, although the effect was much weaker. Overall, lethal outcomes were sometimes intended and sometimes unintended. Violent actions with lethal intent represented instrumental or goal-oriented behavior. The decision to kill a victim reflected tactical concerns, as well as the desire to eliminate rivals, gain retribution, and attain favorable social utilities. However, the research did not rule out the possibility that lethal intent is also affected by frustration or some other innate, nonrational process. The research also did not determine whether all violence is instrumental. Tables, footnotes, and 32 references (Author abstract modified)