Background: cognitive decline and muscle weakness are prevalent health conditions in elderly people. We hypothesised that cognitive decline precedes muscle weakness.Objective: to analyse the temporal relationship between cognitive performance and handgrip strength in oldest old people. Design: prospective population-based 4-year follow-up study.Subjects: a total of 555 subjects, all aged 85 years at baseline, were included into the study.Methods: handgrip strength measured at age 85 and 89 years. Neuropsychological test battery to assess global cognitive performance, attention, processing speed and memory at baseline and repeated at age 89 years. Associations between handgrip strength and cognitive performance were analysed by repeated linear regression analysis adjusted for common confounders. Results: at age 85 and 89 years, better cognitive performance was associated with higher handgrip strength (all, P < 0.03), except for attention. There was no longitudinal association between baseline handgrip strength and cognitive decline (all, P > 0.10), except for global cognitive performance (P = 0.007). Better cognitive performance at age 85 years was associated with slower decline in handgrip strength (all, P < 0.01) after adjustment for common confounders.Conclusion: baseline cognitive performance was associated with decline in handgrip strength, whereas baseline handgrip strength was not associated with cognitive decline. Our results suggest that cognitive decline precedes the onset of muscle weakness in oldest old people.