Normal mammalian cells have a limited lifespan in culture and hypotheses explaining cellular senescence usually fall into one of two categories. One of these postulates that random errors or damage accumulate in essential macromolecules and eventually outstrip the cell's capacity for resynthesis and repair. The second considers the changes when immortal clones are produced from normal cells and in particular the lifespans of hybrids when cells of differing growth potentials are fused. These data can be explained by postulating that the mortal phenotype is dominant and that trans-acting growth inhibitors are involved in limiting lifespan. But the results do not indicate if the inhibitors are the primary cause of senescence or a secondary effect induced by quite different initial events. We suggest that normal cells possess proof-reading mechanisms which monitor the accuracy of chromosome segregation and replication and which can induce the synthesis of growth inhibitors when they detect major errors in chromosome metabolism. It is further postulated that random damage accumulates during the growth of normal cells and eventually leads to detectable chromosome changes and the synthesis of inhibitors. Our hypothesis predicts that the emergence of immortal clones will be linked to the absence of active inhibitors and therefore to a loss in the fidelity of chromosome metabolism. Data are quoted which show that in contrast to normal cells, immortal clones have highly irregular karyotypes, amplify segments of their chromosomes, integrate exogenous DNA efficiently, maintain a constant level of 5-methylcytosine residues and have high frequencies of chromosomal aberrations. The mechanism of the proof-reading is unknown, but it may monitor changes in the patterns by which chromosome domains are attached to the nuclear matrix.