BACKGROUND: This study describes differences in trajectories of self-reported mental health in an ageing cohort, according to their housing, while controlling for confounders. METHODS: The General Health Questionnaire was measured on six occasions as part of Whitehall II cohort study of office-based British civil servants (1985-2009); 10,308 men and women aged 35-55 at baseline. RESULTS: Home-ownership was the predominant tenure at baseline and increased over the life-course, but the social gradient remained. In the bivariate analysis, by phase nine, renters had higher (poorer mental health) GHQ scores (55.48) than owner occupiers (51.98). Those who reported difficulty paying bills or problems with housing had higher GHQ scores at baseline (financial difficulties 57.70 vs 54.34; house problems 58.06 vs 53.99) and this relative difference increased by phase nine (financial difficulties 59.64 vs 51.67; house problems 56.68 vs 51.22). In multivariate models, the relative differences in GHQ scores by tenure increased with age, but were no longer significant after adjusting for confounders. Whereas GHQ scores for those with housing problems and financial difficulties were still significantly higher as participants grew older. CONCLUSION: The social gradient in the effect of home ownership on mental health, which is evident at baseline, diminishes as people get older, whereas housing quality and financial problems become relatively more important in explaining older people's health. Inequalities in housing quality and ability to deal with household financial problems will become increasingly important mental health issues as the population ages.