Cross-national research has increased in the past few decades, resulting in a large body of empirical research. In particular, cross-national studies are often limited in data sources, which restrict variable selection to debatable proxy indicators. This study therefore uses meta-analytic techniques to examine major cross-national predictors of homicide to determine strengths and weaknesses in theory and design. The findings indicate several critical limitations to cross-national research, including biased sample composition, a lack of theoretical clarity in predictor operationalizations, and an overwhelming reliance on cross-sectional design. The predictors that showed the strongest mean effects were Latin American regional dummy variables, income inequality indicators and the Decommodification Index. Conversely, static population indicators, democracy indices, and measures of economic development had the weakest effects on homicide. (Published Abstract)