Endonuclease VIII, a novel presumptive DNA repair enzyme, was isolated from Escherichia coli by FPLC1 purification. The enzyme was found in strains that contained or lacked endonuclease III and was purified by radial flow S-Sepharose, Mono S, phenyl-Superose, and Superose 12 FPLC. Examination of the properties of endonuclease VIII showed it to have many similarities to endonuclease III. DNA containing thymine glycol, dihydrothymine, beta-ureidoisobutyric acid, urea residues, or AP sites was incised by the enzyme; however, DNA containing reduced AP sites was not. HPLC analysis of the products formed by exhaustive enzymatic digestion of damage-containing DNA showed that endonuclease VIII released thymine glycol and dihydrothymine as free bases. Taken together, these data suggest that endonuclease VIII contains both N-glycosylase and AP lyase activities. Consistent with this idea, DNA containing AP sites or thymine glycols, that was enzymatically nicked by endonuclease VIII was not a good substrate for E. coli DNA polymerase I, suggesting that endonuclease VIII nicks damage-containing DNA on the 3' side of the lesion. Also, since monophosphates were not released after treating thymine glycol-containing DNA with endonuclease VIII, the enzyme does not appear to have exonuclease activity. The enzyme activity was maximal in 75 mM NaCl or 5 mM MgCl2. Analysis of endonuclease VIII by both Superose FPLC and Sephadex yielded native molecular masses of 28,000 and 30,000 Da, respectively. SDS-PAGE, in conjunction with activity gel analysis, gave a molecular mass of about 29,000 Da. Furthermore, renaturation of the putative active band from SDS-PAGE gave rise to an active enzyme.