Ataxia-telangiectasia (A-T) is a human disease characterized by high cancer risk, immune defects, radiation sensitivity, and genetic instability. Although A-T homozygotes are rare, the A-T gene may play a role in sporadic breast cancer and other common cancers. Abnormalities of DNA repair, genetic recombination, chromatin structure, and cell cycle checkpoint control have been proposed as the underlying defect in A-T; however, previous models cannot satisfactorily explain the pleiotropic A-T phenotype. Two recent observations help clarify the molecular pathology of A-T: (a) inappropriate p53-mediated apoptosis is the major cause of death in A-T cells irradiated in culture; and (b) ATM, the putative gene for A-T, has extensive homology to several cell cycle checkpoint genes from other organisms. Building on these new observations, a comprehensive model is presented in which the ATM gene plays a crucial role in a signal transduction network that activates multiple cellular functions in response to DNA damage. In this Damage Surveillance Network model, there is no intrinsic defect in the machinery of DNA repair in A-T homozygotes, but their lack of a functional ATM gene results in an inability to: (a) halt at multiple cell cycle checkpoints in response to DNA damage; (b) activate damage-inducible DNA repair; and (c) prevent the triggering of programmed cell death by spontaneous and induced DNA damage. Absence of damage-sensitive cell cycle checkpoints and damage-induced repair disrupts immune gene rearrangements and leads to genetic instability and cancer. Triggering of apoptosis by otherwise nonlethal DNA damage is primarily responsible for the radiation sensitivity of A-T homozygotes and results in an ongoing loss of cells, leading to cerebellar ataxia and neurological deterioration, as well as thymic atrophy, lymphocytopenia, and a paucity of germ cells. Experimental evidence supporting the Damage Surveillance Network model is summarized, followed by a discussion of how defects in the ATM-dependent signal transduction network might account for the A-T phenotype and what insights this new understanding of A-T can offer regarding DNA damage response networks, genomic instability, and cancer.