Considerable efforts have been made to understand the role of oxidative stress in age-related diseases and ageing. The mitochondrial free radical theory of ageing, which proposes that damage to mitochondrial DNA (mtDNA) and other macromolecules caused by the production of reactive oxygen species (ROS) during cellular respiration drives ageing, has for a long time been the central hypothesis in the field. However, in contrast with this theory, evidence from an increasing number of experimental studies has suggested that mtDNA mutations may be generated by replication errors rather than by accumulated oxidative damage. Furthermore, interventions to modulate ROS levels in humans and animal models have not produced consistent results in terms of delaying disease progression and extending lifespan. A number of recent experimental findings strongly question the mitochondrial free radical theory of ageing, leading to the emergence of new theories of how age-associated mitochondrial dysfunction may lead to ageing. These new hypotheses are mainly based on the underlying notion that, despite their deleterious role, ROS are essential signalling molecules that mediate stress responses in general and the stress response to age-dependent damage in particular. This novel view of ROS roles has a clear impact on the interpretation of studies in which antioxidants have been used to treat human age-related diseases commonly linked to oxidative stress.