Socio-economic differences in self-reported disability are well described but much less is known about their associations with more objective measures of physical capacity. The aim was to study socio-economic differences in performance-based physical capacity in 75-year-old persons, examining changes in performance at five- and ten-year follow-up intervals. At the baseline 350 residents of the city of Jyväskylä, Finland, aged 75 were interviewed and 295 of them took part in clinical examinations. The corresponding figures at the five-year follow-up were 234 and 191 and at the ten-year follow-up 139 and 103. The statistical significance of differences in physical capacity between the socio-economic groups and genders were tested using ANOVA in univariate and repeated measures models and ANCOVA, with confounders added to the models. Generally, higher education and income were separately related to better maximal walking speed and vital capacity at every measurement point. In addition, higher income was related to better maximal isometric hand grip strength at both follow-ups. When education and income were in the same model, only income was related to physical capacity, almost without exception. Similarly, in the five- and ten-year follow-up periods, both education and income groups showed a parallel decline in physical capacity. The association between income and physical capacity remained even after adjusting for smoking, physical activity and number of chronic diseases. The results indicate that elderly people in disadvantaged socio-economic groups show lower levels of performance in almost all domains of physical capacity, but change in capacity over time does not differ significantly between either markers of socio-economic position.