The NVDRS integrates death certificate, coroner/medical examiner, and law enforcement data. The article found that overall, data collected across multiple source documents matched at greater than 70 percent; however inconsistencies have implications for analyzing data from systems with multiple source documents. It was advised that the understanding and mitigation of data mismatches would increase the consistency of data on which violence prevention programs are developed. Data from the 2003-2004 National Violent Death Reporting System were used to compare consistency of homicide variables across multiple source documents. Included in this analysis were 5,737 homicide incidents. Variables included victim demographics, manner of death, autopsy and pregnancy status, place, date and location of injury/death, and suspected use of alcohol. Demographic variables matched from lows of 71 percent for marital status to 99.9 percent for race. Injury/death variables matched from over 72 percent for date of injury to over 99 percent for state of injury. Situational variables ranged from 75 percent for suspected alcohol use to 97 percent for pregnancy status. Tables, references