Like ageing insects, worms and mammals, growth-arrested Escherichia coli cells accumulate oxidatively damaged proteins. In the early stages of the E. coli stationary phase, this oxidation is caused by an increased production of aberrant proteins, which are especially susceptible to oxidative attack. This route of oxidation appears to elude the classical oxidative defence proteins. The failure of growth-arrested cells fully to combat oxidative damage may also be linked to a trade-off between proliferation activities (primarily directed by the housekeeping sigma factor, sigma70) and maintenance (primarily directed by sigmaS). This trade-off is regulated by the alarmone ppGpp such that elevated ppGpp levels allow sigmaS, and other alternative sigma factors, to work in concert with sigma70 by shifting their relative competitiveness for RNA polymerase binding. However, even during elevated ppGpp levels and stasis, E. coli cells maintain a basal transcription of housekeeping sigma70-dependent genes, and resources are thus partly diverted from maintenance and stress defences to activities relating to proliferation. An alternative view argues for ppGpp being involved in programmed cell death upon growth arrest by regulating chromosomally located toxin-antitoxin loci. Thus, models of bacterial senescence, like those dealing with ageing in higher organisms, encompass both stochastic deterioration theories and programming theories. This review summarizes and evaluates these models.