This study compared data from death certificates (DC), medical examiner (ME) reports, and Uniform Crime Reporting (UCR) Program for homicides among children age 5 years or younger in Jefferson County, Alabama between 1988 and 1998. Records from each source were matched independently to records from the other two sources. Kappa coefficients measured the chance-corrected agreement between sources. The strongest correlations were for victim's gender and race. Correlations were weakest for victim's relationship to the offender and offender's gender. DC and ME data demonstrated excellent agreement and provided a comprehensive count of homicides in Jefferson County, whereas UCR comparisons had lower correlations. However, the UCR is a valuable source of child homicide data in the United States that is underutilized in public health research.