Crime rates dropped unexpectedly and dramatically in the 1990s in Canada and the USA. The decline was not restricted to particular types of crime, the particular methodologies of crime reports, demographic characteristics, or geographical areas. Psychological studies of individuals have suggested a link between crime and different types of risky behavior (e.g., dangerous driving, unsafe sex, substance use). Based on this link, we examined whether national rates of various risky behaviors declined in the 1990s, and whether rates of crime and risky behavior covary over time. Several American and Canadian databases reporting annual or biennial data on risky behavior indicators were examined. Results indicate that most of the risky behavior indicators in the domains of violent behavior, accidents, sexual behavior, and school dropout declined in the 1990s. Furthermore, time series analyses suggest that rates of various risky behaviors tend to covary with homicide over long periods of time. An important exception to these results is substance use in various contexts. We discuss some theoretical implications of the results.