Cross-national homicide researchers have traditionally relied on homicide data drawn from criminal justice records, medical records, or some combination of the two. However, there have been few studies that directly examine the methodological differences between homicide data sources and the consequences of these differences. In the current study we examine the gap in homicide rates for data compiled by the UNODC and WHO. We provide a set of theoretically grounded hypotheses aimed at explaining differences between criminal justice and medical sources of homicide data. Analyzing 1,597 cases for 82 countries from 1991 to 2020, results from ANOVA and Wilcoxon signed-rank tests indicate that countries with a great deal of homicides have larger gaps than those with relatively few homicides. According to our quantile regression analysis for panel data models, the divide between criminal justice and medical sources of homicide data declines as medical and criminal justice systems grow more efficient. We also find that the gap in homicide rates narrows as countries globalize. Average marginal effects analysis indicates that health quality exerts the greatest reductions in the homicide gap. These findings remain robust when we use dynamic panel model as a robustness check. Implications for theory and policy are discussed.