Of the two methods used (degree of allele sharing and a kinship matching approach), profile matching based on the degree of allele sharing proved to be an inefficient means of establishing familial matches. There was little correlation between the degree of allele sharing and the number of hits required to find a true match. Use of this method to match crime-scene profiles to a forensic database of any reasonable size would result in numerous false matches between common alleles. Furthermore, allele-matching patterns between siblings are not always informative. The percentage of loci where single alleles matched was found to be equal for both known siblings and unrelated pairs in a short tandem repeat (STR) system that included the 13 core loci of the Combined DNA Index System (CODIS). Kinship matching was much more efficient at identifying sibling pairs, with the caveat that relatively high sibship indexes were required to locate a true match. Two sample populations were created for this study. The first dataset consisted of DNA profiles from samples used in previously tested sibship cases. A second larger set of data simulating an offender database consisted of DNA profiles from samples used in previously tested parentage cases. In addition to these 2 populations, 12 profiles from individuals unrelated to members of either database were used as nonmatching controls. 2 figures and 11 references