For both the UCR, which measures crime reported to police agencies throughout the country, and the NCS, which measures victimization rates for the nation, sample evidence exists to question their validity. Although no 'true' crime rate is known against which to measure the validity of UCR and NCS, empirical comparisons can be made which assess the validity of a measure. The procedures used in this study are called 'convergent' validation (Campbell and Fiske, 1959) and 'nomological' validation (Selltiz et al., 1976). Convergent validation is based on the argument that two variables which validly measure the same concept should correlate highly. Nomological validation is based on the argument that two variables not measuring the same concept which theory suggests are related should correlate highly if both are valid measures; for example, if median income of cities is held to be correlated with crime rates, there should be a correlation between the median income of cities and the UCR and NCS rates across cities. With respect to the issue of the degree to which the two measures of crime correlate across cities, only three of the six offenses examined (motor vehicle theft, robbery, burglary, personal larceny, rape, and aggravated assault) showed even a moderate degree of convergent validity: motor vehicle theft, robbery, and burglary. The nomological validity of the two measures was examined by correlating each measure with a number of urban structural characteristics thought to be correlated with the true rate of crime. This analysis showed that overall, the NCS had a greater degree of nomological validity. Tabular data and 37 references are provided.