The value of the HLA-A,B system in excluding a man falsely accused of paternity, the power of exclusion (A), was examined by four different methods. Methods to define A were: analysis of artificially created false trios, calculation of the proportion of all exclusion cases (corrected for unexcluded men who were not fathers) in which the HLA system showed evidence of an exclusion, derivation from individual Random Man Not Excluded (RMNE) values, and derivation from individual Paternity Index (PI) values. The false trios were limited within the same ethnic group (whites and blacks) so that no mixed ethnic matings were considered. It was assumed that all of the men in the artificial trios were not fathers. Twenty-one percent of the observed exclusions were based on A locus markers, 36 percent on B locus markers, and 43 percent on both A and B markers. Among 510 false white trios, the man was excluded in 476 cases, yielding a power of exclusion (A) for whites of 93.3 percent. One hundred thirty men were excluded in 140 black artificial trios, giving a value of 92.9 percent for A in blacks. Similar values, 94.4 percent for whites and 93.6 percent for blacks, were derived from the proportion of all exclusions detected by the HLA system. A was also derived by computer analysis on the same cases with formulas that use RMNE and PI values. These results also agreed. This concordance by all four methods enhanced the validity of the A values of about 93 to 94 percent for both whites and blacks.(ABSTRACT TRUNCATED AT 250 WORDS)