Despite numerous studies, the nature of the unemployment-crime relationship remains controversial. The relationship should be clearer for some segments of the population than for others, but is obscured by the use of general population data. Exploring this possibility through the use of a model developed by Cantor and Land (1985), a time-series analysis is conducted to determine relationships among age- and race-specific rates of unemployment and corresponding rates of arrests for homicide, robbery, and burglary for the United States during the period 1959–1987. Negative criminal opportunity-related and positive criminal motivation-related effects are found at the aggregate level, but these vary among age groups and are more evident for white than for African American arrest rates. Further, these effects hold even when controlling for the potential influence of other variables identified in recent research as having an impact on the unemployment-crime relationship.