There are considerable differences in the timing, type and extent of age-related decline between individuals who share the same chronological age, which may be driven by a combination of genetic, stochastic and environmental factors. Biomarkers of ageing (BoA) that can discriminate between individuals who differ in their biological age will therefore be useful to understand biological mechanisms, develop and test interventions and allow the prediction of age-related events so interventions can be implemented. In recent years, a variety of mechanistic candidate BoA have been discovered on the basis of a greatly improved understanding of the cellular and molecular biology of ageing. These include various measures of oxidative stress, which is thought to contribute causally to the ageing of organisms via its acceleration of cellular senescence. However, their reliability and validity as BoA, especially within population based cohorts are scarce. This study therefore focused on various oxidative stress-related measures as candidate BoA including: reactive oxygen species (ROS) production from dysfunctional mitochondria, by measuring superoxide levels, mitochondrial mass and mitochondrial membrane potential in blood mononuclear cells by flow cytometry; and also markers of lipid peroxidation, F2-isoprostanes, by measuring 8-iso Prostaglandin F2 by Automated Dissociation Enhanced Lanthanide Fluorescence Immunoassay (AutoDELFIA). Despite providing evidence of experimental reliability for all measures and also some evidence of construct validity for ROS production from dysfunctional mitochondria in terms of: associations with chronological age, associations with some markers of oxidative stressinduced cellular senescence, validation in a dietary restricted animal model of ageing and a role in an immunosenescent phenotype; there was no evidence of predictive validity in terms of longevity or age-related health outcomes in a population based cohort of the very old, the Newcastle 85+ study. This questions the predictive validity of these parameters as candidate BoA in the very old population.