The relevance of using dominant random amplified polymorphic DNA (RAPD) fingerprints for estimating population differentiation was investigated when typically small population sample sizes were used. Haploid sexual tissues were first used to determine genotypes at RAPD loci for 75 eastern white pines (Pinus strobus L.) representing five populations. Dominant RAPD fingerprints were then inferred from genotypic data for each individual at each locus, and gene diversity estimates from both sources of data were compared. Genotypic information at RAPD loci indicated little or no differentiation among populations, similar to allozyme loci. However, estimates of population differentiation derived from dominant RAPD fingerprints according to various common methods of analysis were generally inflated, especially when all fragments were considered. Simulations showed that an increase in loci sampling and population sample sizes did not significantly alleviate the biases observed.