The non-homologous end joining (NHEJ) pathway is used in diverse species to repair chromosome breaks, and is defined in part by a requirement for Ku. We previously demonstrated mammalian Ku has intrinsic 5′ deoxyribosephosphate (5′dRP) and apurinic/apyrimidinic (AP) lyase activity, and showed this activity is important for excising abasic site damage from ends. Here we employ systematic mutagenesis to clarify the protein requirements for this activity. We identify lysine 31 in the 70 kD subunit (Ku70 K31) as the primary candidate nucleophile required for catalysis, but additional mutation of Ku70 K160 and six other lysines within Ku80 were required to eliminate all activity. Ku fromSaccharomyces cerevisiaealso possesses 5′dRP/AP lyase activity, and robust activity was also reliant on lysines in Ku70 analogous to K31 and K160. By comparison, these lysines are not conserved inXenopus laevisKu, and Ku from this species has negligible activity. A role for residues flanking Ku70 K31 in expanding the range of abasic site contexts that can be used as substrate was also identified. Our results suggest an active site well located to provide the substrate specificity required for its biological role.