Pituitary growth hormone (GH), like several other protein hormones, shows an unusual episodic pattern of molecular evolution in which sustained bursts of rapid change are imposed on long periods of very slow evolution (near-stasis). A marked period of rapid change occurred in the evolution of GH in primates or a primate ancestor, and gave rise to the species specificity that is characteristic of human GH. We have defined more precisely the position of this burst by cloning and sequencing the GH genes for a prosimian, the slow loris (Nycticebus pygmaeus) and a New World monkey, marmoset (Callithrix jacchus). Slow loris GH is very similar in sequence to pig GH, demonstrating that the period of rapid change occurred during primate evolution, after the separation of lines leading to prosimians and higher primates. The putative marmoset GH is similar in sequence to human GH, demonstrating that the accelerated evolution occurred before divergence of New World monkeys and Old World monkeys/apes. The burst of change was confined largely to coding sequence for mature GH, and is not marked in other components of the gene sequence including signal peptide, 5' upstream region and introns. A number of factors support the idea that this episode of rapid change was due to positive adaptive selection. Thus (1) there is no apparent loss of function of GH in man compared with non-primates, (2) after the episode of rapid change the rate of evolution fell towards the slow basal level that is seen for most mammalian GHs, (3) the accelerated rate of substitution for the exons of the GH gene significantly exceeds that for introns, and (4) the amino acids contributing to the hydrophobic core of GH are strongly conserved when higher primate and other GH sequences are compared, and for coding sequences other than that coding for hydrophobic core residues the rate of substitution for non-synonymous sites (K(A)) is significantly greater than that for synonymous sites (K(S)). In slow loris, as in most non-primate mammals, there is no evidence for duplication of the GH gene, but in marmoset, as in rhesus monkey and man, the putative GH gene is one of a cluster of closely related genes.