Divergence dates among primates were estimated by molecular clock analysis of DNA sequence data. A molecular clock of eta-globin pseudogene was calibrated by setting the date of divergence between Catarrhini and Platyrrhini at 38 million years (Myr) ago. The clock gave dates of 25.3 +/- 2.4, 11.9 +/- 1.7, 5.9 +/- 1.2, and 4.9 +/- 1.2 Myr ago ( +/- refers to standard error) for the separation of rhesus monkey, orangutan, gorilla, and chimpanzee, respectively, from the line leading to humans. In placing confidence intervals of the estimates in a robust way, a bootstrap method was used. The 95% confidence intervals are 20.5-29.5, 9.0-14.8, 4.1-7.8, and 3.1-7.0 Myr ago for the separation of rhesus monkey, orangutan, gorilla, and chimpanzee, respectively. By a molecular clock dating of the Prosimii-Anthropoidea splitting, it was suggested that the evolutionary rate of the eta-globin gene was high early in primate evolution and subsequently decreased in the line of Anthropoidea. And, by a relative rate test using bootstrap sampling, the possibility of further decrease of the rate (more than 10%) in the line of Hominoidea compared with that of Cercopithecoidea was suggested. Therefore, the above dating of the splittings within Hominoidea may be biased slightly toward younger dates. On the other hand, mitochondrial DNA (mtDNA) seems to have evolved in mammals with a more uniform rate than the eta-globin gene. The ratio of the dates of orangutan splitting to chimpanzee splitting is larger for the mtDNA clock than that for the eta-globin clock, suggesting the possibilities of mtDNA introgression among the early hominids and the early African apes, and/or of mtDNA polymorphism within the common ancestral species of orangutan and the African apes that obscures the date of the true species separation of orangutans.