The study found that by using crime rates and arrest statistics as a basis for analysis, the offense rates for males of both races and all important age groups increased in each city by a considerable margin. A small proportion of the increase in crime could have been predicted by assuming that age- and race-specific crime rates would continue at 1960 levels, but the age and race structure of the cities would have changed in the pattern that has been experienced. This method is called 'prospective demography,' even though accurate use of the method in 1960 would have required foreknowledge of the characteristics of 1970 urban populations. Using what is termed 'retrospective demography,' i.e., assuming the rates of crime would change for each age and racial group as they did, the differences in population composition explains a much larger share of the noted increase in crime rates; it cannot be projected to future time periods without assuming age-, race-, and sex-specific crime rates. Variance between cities was substantial, so much so that generalization from aggregate national data seems less justified as a result of this exercise. There is some evidence that robbery is more concentrated among young offenders (15-19) in 1970 than in 1960 and strong evidence that auto theft is less concentrated among the young in the later year. There is no clear pattern over time for burglary and homicide. Some evidence indicates that robbery is more concentrated among black offenders in 1970 than in 1960. This is probably related to the relatively high concentration of young males in the black population. The racial concentration of homicide is stable in three of the five cities and increases in the other two. Tabular data are appended.