In the study of molecular evolution, it is important to know the number of nucleotide substitutions per site (d) between DNA sequences. There are many different methods for estimating this number (for reviews, see Nei 1987, chap. 5; Gojobori et al. 1990)) but all of them depend on some simplifying assumptions and do not always give reliable estimates. Two important factors that should be considered in the estimation of d are the inequality of the rates of transitional and transversional nucleotide substitution (transition-transversion bias) and the deviation of the G+ C content (0) from 0.5 (G+ C-content bias).

Kimura (1980) developed a method of estimating d for the case where the transition-transversion bias exists. In his method, however, the frequencies of the four nucleotides A, T, C, and G are all assumed to be equal to 0.25, throughout the evolutionary time. In practice, this assumption usually does not hold. For example, the G+ C content at the third-codon positions of the coding region of Drosophila mitochondrial DNA (mtDNA) is-GO. 1 (Clary and Wolstenholme 1985). It is therefore useful to develop a method that will take care of both the transition-transversion and G+ C-content biases. In the following, I present one such method and compare the method’s reliability with that of others.