This paper presents exploratory analysis of these competing assertions. Three aspects of the relationships between crime, youth, and punishment were investigated. A time-series model of crime rates, punishment risk, and size of the youth population was estimated using standard econometric methods. Alternative specifications of model form and punishment risk consistently supported the view that changes in punishment risks, rather than changes in the size of the youth population, account for changes in crime trends. An analysis of victim responses in the National Crime Survey adjusted estimates of the number of crimes committed by various age groups to account for the possibility that youth gang participation distorted perceptions of the prevalence of youth in criminal activity. It found that youth were indeed the most criminally active even after adjustment for group crimes. This analysis suggests that age composition of the population also may play some role in crime trends. A simple age composition model was developed to project crime trends based on age factors alone. After fixing the crime incidence rate for each age group at 1983 rates, the model forecast crimes for various points between 1960 and 1983. These forecasts diverged rapidly from actual crime counts, suggesting that age trends alone are inadequate predictors of crime trends. 5 tables and 15 references. (Author abstract modified)