Testing structural and cultural explanations of the cross-sectional variation of homicide rates is an increasingly popular criminological pursuit and has become easier over time given increasing accessibility of crime and explanatory data at different levels of analysis. One recent trend is the use of the county as the unit of analysis with evidence of solid methodological and substantive research in this area. However, its popularity has raised concern with researchers arguing the necessity of undertaking analyses at the county-level and employing county-level crime and homicide data without recognizing the limitations on their usefulness for model estimation. The use of counties can result in an excess of zero counts for homicide rates due to the rarity of serious crimes and the relatively small populations of the more than 3,000 counties in the United States. Using counties as the unit of analysis can result in skewed distributions, imprecise estimates that vary by county population and in some cases a high proportion of zero counts. Selecting the unit of analysis depends first and foremost on theory; however, no matter what the unit or the methodology, all tests depend on trustworthy data, and just because data are readily available does not mean they are appropriate for consumption. Table, notes, references