Data were collected on 175 homicides in Detroit between January 1999 and December 2002. Among the variables coded for these homicides were gang involvement, apparent motive for the homicide, and level of drug involvement. The dependent variable was the drug-relatedness of the homicide. Factors taken into account in this assessment were the presence of drugs at the scene, whether the victim or offender was buying or selling drugs at the time of the killing, and whether the killing stemmed from competition with a rival drug dealer. These factors are similar to those used in Goldstein's analytical scheme for the drug-homicide link, but with the exception of the psychopharmacological effects of drugs in homicidal behavior. Although this factor is included in Goldstein's model, it was excluded from the current study because of the difficulty of measuring such a variable. A multinomial logistic regression technique was used to determine the existence and type of drug-related links to each homicide. Approximately 50 percent of the homicides did not involve any type of drug connection. An unanticipated finding was the negligible role that drug markets apparently had in predicting different types of homicide events. Those homicides in which drugs played a part involved younger offenders and victims. The relationship between victim age and peripheral drug events suggests a lifestyle effect in which youth are involved in partying, drug use, and spending time with fellow drug users who are likely to carry firearms. This is apparently a volatile lifestyle that poses a high risk for homicides. Future research and policy implications are discussed. 3 tables, 6 notes, 29 references, and appended examples of drug-related homicide types