Studies of human genetic diseases have suggested a higher mutation rate in males than in females and the male-to-female ratio (alpha) of mutation rate has been estimated from DNA sequence and microsatellite data to be about 4-6 in higher primates. Two recent studies, however, claim that alpha is only about 2 in humans. This is even smaller than the estimates (alpha > 4) for carnivores and birds; humans should have a higher alpha than carnivores and birds because of a longer generation time and a larger sex difference in the number of germ cell cycles. To resolve this issue, we sequenced a noncoding fragment on Y of about 10.4 kilobases (kb) and a homologous region on chromosome 3 in humans, greater apes, and lesser apes. Here we show that our estimate of alpha from the internal branches of the phylogeny is 5.25 (95% confidence interval (CI) 2.44 to infinity), similar to the previous estimates, but significantly higher than the two recent ones. In contrast, for the external (short, species-specific) branches, alpha is only 2.23 (95% CI: 1.47-3.84). We suggest that closely related species are not suitable for estimating alpha, because of ancient polymorphism and other factors. Moreover, we provide an explanation for the small estimate of alpha in a previous study. Our study reinstates a high alpha in hominoids and supports the view that DNA replication errors are the primary source of germline mutation.