Principal components analysis was used to evaluate finger ridge–count variability as an indicator of genetic relationships between populations. The analysis was carried out on American White, American Black and African Black samples, each including both sexes. Each individual is represented as a vector of 20 counts, a radial and an ulnar count for each digit. No assumptions were made prior to analysis concerning the number of meaningful components, and all were examined sequentially. The first five eigenvectors extracted from the within-group correlation matrix have loadings very similar to those previously described by Roberts and Coope ('75). However, it is the component scores derived from the sixth eigenvector which show the most marked variation, accounting for 45% or more of the D2in all Black-White comparisons. A number of other components also show significant intergroup heterogeneity, but they often do not accord with what is known of the genetic relationships between the populations. Apparently a large amount of ridge-count variation is not genetically meaningful, at least as far as these populations are concerned.