The findings suggest that although the decline in homicide rates may partly reflect a decrease in the amount of drug-market activity and/or the zealousness of drug law enforcement, it also may be linked to a weakening relationship between drug market activity and violence. There was little evidence that the drug market-homicide link was moderated by the age of a city's drug market. Support was fairly limited for the theory that within-city change in structural disadvantage impacted the strength of the time-period variation in the relationship between the drug market and homicide rates. The findings suggest that the focus on the variation in the impact of drug markets on violence over time is too narrow. Future studies should develop and test conceptual models that assist in explaining when and why the impact of various large-scale predictors of homicide rates may be altered. The study collected data that repeatedly measured homicide rates, drug markets, and sociodemographic characteristics for a sample of 132 U.S. cities with a minimum total population of 100,000 persons in the 1980, 1990, and 2000 decennial census years and a minimum of 12 out of 17 years of data for the dependent variable (homicide rate per 100,000 persons) and the key drug market measure (arrest rate for the sale, distribution, or manufacture of cocaine, opiates, or derivatives) during the 1984-2000 period. 2 tables, 89 references, and appendix