The long-term decline in homicide rates seems to coincide with a disproportionate decline in elite homicide and a drop in male-to-male conflicts in public space. A range of explanations for this decline are offered, including the effects of the civilizing process, strengthening state powers, the Protestant Reformation, and modern individualism. Understanding of the accuracy and comparability of historical estimates of homicide rates should be improved. Some promising strategies on improving estimates of homicide rates are capture-recapture methods, better population estimates, or more broadly based information on the effects of improved medical technology. As knowledge of overall levels of lethal violence increases, it may become more important to examine developments in subtypes such as family homicide, infanticide, or robbery-related killings. Existing research has not yet fully explored historical variation in contextual variables. Examining to a full extent quantifiable information about offenders, victims, and situations - possibly using standardized tools across studies - may significantly contribute to knowledge. It would also be useful to improve understanding of the geo-historical map of homicide in Europe. France and Spain are conspicuously missing, and more information about trends in Italy and different areas in the German-speaking parts of Europe would enrich comparative analyses. Further empirical research could benefit from a more coherent set of theoretically based questions. It might be beneficial to adopt systematically comparative perspectives in future research. Findings from social history research may provide indicators of historical and geographic variation in patterns of formal social control, levels of literacy, political conflict, and the commercialization of the economy. By comparing regions that systematically differ in these respects, more could be learned about what variables contribute to changing levels of homicide. 6 tables, 10 figures, 182 references