The concept focuses on individual points rather than on areas. The population potential for a specific point considers the people surrounding a particular point in terms of both their numbers and their proximity. The crime risk is the product of both the numbers and the proximity of crime incidents. These concepts were applied using 1970 police and census data for 703 census tracts in Los Angeles. Predictive equations were developed and tested using 20 points taken at random in the study area. Safe points were those farthest from all human populations, regardless of race. Within the more populated areas, points surrounded by intact husband-wife couples were less risky than other points. It is possible to make effective risk predictions for all points on the map, subject to specific assumptions. Details of the prediction model, data tables, and 4 reference notes. (Author abstract modified)