A way to estimate the index of nucleotide diversity (pi) from band match frequencies in random amplified polymorphic DNA and amplified fragment length polymorphism data is described. pi is shown to be a simple function of the proportion of mismatched bands between two individuals drawn at random from a population (phi) and the number of discriminating sites in the amplification system. The method is computationally and conceptually simple and avoids some of the assumptions inherent in other approaches: the relationship is independent of the base composition of the target DNA and avoids the bias inherent in estimations of allelic frequencies in dominant systems. Only two individuals from a population are needed to estimate pi. This economy of material suggests utility of this approach in conservation genetics or other fields where obtaining large samples is impractical or undesirable.