DNA polymerase III of the yeast Saccharomyces cerevisiae has been reported to be encoded at the CDC2 locus based on two observations. First, the CDC2 gene has homology to known DNA polymerase genes [Boulet et al. (1989) EMBO J. 8, 1849-1854], and second, the mutants cdc2-1 and cdc2-2 yield little or no DNA polymerase III activity in vitro [Boulet et al. (1989); Sitney et al. (1989) Cell 56, 599-605]. We describe here the isolation of temperature-sensitive DNA polymerase III from cdc2-2 strains. Our results provide direct experimental confirmation of the previously inferred gene/enzyme relationship and verify the conclusion that DNA polymerase III is required to replicate the genome. We isolated DNA polymerase III from two cdc2-2 strains, one containing the wild-type allele for DNA polymerase I (CDC17) and the other a mutant DNA polymerase I allele (cdc17-1). Yields from cdc2-2 cells of both DNA polymerase III activity and an associated 3'-5'-exonuclease activity [exonuclease III; Bauer et al. (1988) J. Biol. Chem. 263, 917-924] were decreased relative to yields from CDC2 cells. DNA polymerase III activity from cdc2-2 strains is thermolabile, displaying at least a 4-fold reduction in half-life at 44 degrees C. The activity is also labile at 37 degrees C, a temperature which is restrictive for growth of cdc2-2 but not CDC2 strains. At 23 degrees C, a temperature which is permissive for growth of both cdc2-2 and CDC2 strains, the mutant and wild-type DNA polymerase III activities display equal stability. These observations provide a demonstrable biochemical basis for the thermosensitive phenotype of cdc2-2 cells.