The missing data addressed include both non-reports (“unit missingness”) and incomplete reports (“item missingness). For incomplete case data--i.e., missing characteristics on victims, offenders, or incidents--the authors introduced a multiple imputation (MI) approach based on a log-linear model for incomplete multivariate categorical data. Next, in order to adjust for “unit missingness,” the authors adopted a weighting scheme linked to FBI annual estimates of homicide counts by State and mortality data from the National Center for Health Statistics related to decedent characteristics in coroners’ reports for deaths classified as homicides. The results of these data revisions are a fully imputed SHR database for 1976-2005. The study compared the joint distributions of several important variables for both the actual cases and the fully imputed dataset. In all instances, the relationships between important variables remained relatively unchanged when using the imputed data. Although there were some slight differences in the distribution of offender race by homicide circumstance and victim-offender relationship by homicide circumstance, these distributions were still similar to what was observed only with the actual cases. In addition, the divergence was expected because of the higher rates of unsolved murder in urban areas. Although there were some slight differences in the distribution of weapon and offender age regarding knives and other weapons, the relationship between offender age and guns remained virtually unchanged. In an examination of the distribution between victim sex and victim-offender relationship, the differences between the actual cases and the fully imputed data were consistent with prior research that suggests an inverse relationship between the closeness of victim-offender relationship and the likelihood of “missingness.” 7 tables, 3 figures and 25 references