Simian virus 40 (SV40) DNA replication requires the coordinated action of multiple biochemical activities intrinsic to the virus-encoded large tumor antigen (T antigen). We report the preliminary biochemical characterization of the T antigens encoded by three SV40 mutants, 5030, 5031, and 5061, each of which have altered residues within or near the ATP binding pocket. All three mutants are defective for viral DNA replication in cultured cell lines. However, while 5030 and 5031 can be complemented in vivo by providing a wild-type T antigen in trans, 5061 exhibits a strong trans-dominant-negative phenotype. In order to determine the basis for their replication defects and to explore the mechanisms of trans dominance, we purified the T antigens encoded by each of these mutants and examined their activities in vitro. The 5061 T antigen had no measurable ATPase activity and failed to hexamerize in response to ATP, and its affinity for the SV40 origin of DNA replication (ori) DNA was not increased by ATP. In contrast, the 5030 and 5031 T antigens exhibited at least some ATPase activity and both readily formed hexamers in the presence of ATP. These mutants differed in that 5030 was very defective in an ori-dependent unwinding assay while 5031 retained significant activity. Both the 5030 and 5031 T antigens bound to ori-containing DNA, but the binding was less efficient than that of wild-type T antigen and was not affected by the presence of ATP. These results suggest that 5030 and 5031 are defective in some aspect of communication between the ATP binding and DNA binding domains and that the ability of ATP to induce T-antigen hexamerization is distinct from its action to increase the affinity for ori. Finally, all three mutants were defective for the ability to support SV40 DNA replication in vitro. Both the 5031 and 5061 T antigens inhibited wild-type-T-antigen-stimulated replication in vitro, while the 5030 T antigen did not. The fact that the 5031 T antigen was trans dominant in the in vitro assays but not in vivo indicates that the in vitro system does not accurately reflect events occurring in vivo.