Victim characteristics measured were sex, race, and prior arrest record; and neighborhood characteristics were determined by median income, unemployment rate, median educational level of residents, and low owner occupancy of housing. Other measured variables related to the homicide were location of the body, the weapon used, the characteristics of the homicide, and community area size. Hierarchical generalized linear modeling (HGLM) was used to gain the benefits of multilevel modeling with a dichotomous dependent variable. The statistically significant variables that decreased the likelihood of a homicide case being cleared by arrest included cases with Latino victims, older victims, homicides that occurred in taverns, homicides committed with firearms, those that were related to other felonies, drug-related or gang-related circumstances, or unknown circumstances. Statistically significant variables that increased the likelihood of a homicide case being cleared by arrest were cases that occurred in homes and cases that occurred in community areas with greater home-ownership rates. The importance of each of these findings is discussed in relation to the hypotheses and previous literature. The study concludes that there is overwhelming support for the importance of factors unrelated to police discretionary decisions in determining homicide clearance rates by arrest; however, further research is needed to examine the variables of Latino victim and areas of greater home ownership as factors in the homicide clearance rate. 1 table, 9 notes, and 54 references