Bovine adenoviruses (BAVs) are important pathogens causing significant economic losses to the cattle industry. We have been interested in the differences among serotypes of these viruses, particularly in their pathogenicity and host range. As part of our efforts to better understand these viruses, we have determined the nucleotide sequences for serotype 3 (BAV3) at map coordinates beween 11.7 and 23.7% and for serotype 2 (BAV2) between 13.1 and 24.0%. Analyses of these sequences revealed large open reading frames (ORFs) encoded within the leftward-reading strand of the viral DNA. The coding capacity of the ORF in BAV3 is 1,167 amino acid residues and 1,138 in BAV2. A search in the GenEMBL protein sequence databank for homology to the predicted polypeptide products of these ORFs established their identity as that for the adenovirus (Ad) DNA polymerase (DNA pol). The deduced polypeptide sequences were aligned with each other and with other known Ad DNA pols to reveal regions of homology and similarity. The comparison at the amino acid sequence level not only showed that the bovine Ad DNA pols from the two serotypes are quite distinct from each other, but also revealed that Ad DNA pols contain multiple domains that are highly conserved among human, canine and bovine Ads. These conserved domains are likely important for the multiple functions attributed to Ad DNA pol, which include catalysis of its own initiation complex, elongation of nascent DNA strand, as well as correction of DNA replication errors.