Between July 1, 1979, and March 15, 1981, there were 22 unsolved homicides and two unsolved disappearances of Atlanta children. Using epidemiologic methods, we attempted to identify factors that had put children at an increased risk of homicide. That all victims in this cluster were black, killed away from home, and that asphyxiation was overrepresented suggests that the cluster was discrete. The cluster was not homogeneous in relation to location of the victim's area of residence or location of the body; however, the median distance of 9.3 miles from home to body suggests that in some cases a motor vehicle was involved. A neighborhood-based study of the male victims and age- and sex-matched controls showed that victims more often ran errands for money (relative risk, 7.9) and were more often alone on the streets or in shopping centers; therefore, they may have been more approachable than other children in the neighborhood.