Cyclin-dependent kinases (CDKs) function as central regulators of both the cell cycle and transcription. CDK activation depends on phosphorylation by a CDK-activating kinase (CAK). Different CAKs have been identified in budding yeast, fission yeast, and metazoans. All known CAKs belong to the extended CDK family. The sole budding yeast CAK, CAK1, and one of the two CAKs in fission yeast, csk1, have diverged considerably from other CDKs. Cell cycle regulatory components have been largely conserved in eukaryotes; however, orthologs of neither CAK1 nor csk1 have been identified in other species to date. To determine the evolutionary relationships of yeast and metazoan CAKs, we performed a phylogenetic analysis of the extended CDK family in budding yeast, fission yeast, humans, the fruit fly Drosophila melanogaster, and the nematode Caenorhabditis elegans. We observed that there were 10 clades for CDK-related genes, of which seven appeared ancestral, containing both yeast and metazoan genes. The four clades that contain CDKs that regulate transcription by phosphorylating the carboxyl-terminal domain (CTD) of RNA Polymerase II generally have only a single orthologous gene in each species of yeast and metazoans. In contrast, the ancestral cell cycle CDK (analogous to budding yeast CDC28) gave rise to a number of genes in metazoans, as did the ancestor of budding yeast PHO85. One ancestral clade is unique in that there are fission yeast and metazoan members, but there is no budding yeast ortholog, suggesting that it was lost subsequent to evolutionary divergence. Interestingly, CAK1 and csk1 branch together with high bootstrap support values. We used both the relative apparent synapomorphy analysis (RASA) method in combination with the S-F method of sampling reduced character sets and gamma-corrected distance methods to confirm that the CAK1/csk1 association was not an artifact of long-branch attraction. This result suggests that CAK1 and csk1 are orthologs and that a central aspect of CAK regulation has been conserved in budding and fission yeast. Although there are metazoan CDK-family members for which we could not define ancestral lineage, our analysis failed to identify metazoan CAK1/csk1 orthologs, suggesting that if the CAK1/csk1 gene existed in the metazoan ancestor, it has not been conserved.