INTRODUCTION: Physical function and cognition seem to be interrelated, especially in the oldest-old. However, the temporal order in which they are related and the role of brain health remain uncertain. METHODS: We included 338 participants (mean age 93.1 years) from two longitudinal cohorts: the UCI 90+ Study and EMIF-AD 90+ Study. We tested the association between physical function (Short Physical Performance Battery, gait speed, and handgrip strength) at baseline with cognitive decline (MMSE, memory tests, animal fluency, Trail Making Test (TMT-) A, and digit span backward) and the association between cognition at baseline with physical decline (mean follow-up 3.3 years). We also tested whether measures for brain health (hippocampal, white matter lesion, and gray matter volume) were related to physical function and cognition and whether brain health was a common driver of the association between physical function and cognition by adding it as confounder (if applicable). RESULTS: Better performance on all physical tests at baseline was associated with less decline on MMSE, memory, and TMT-A. Conversely, fewer associations were significant, but better scores on memory, TMT-A, and digit span backward were associated with less physical decline. When adding measures for brain health as confounder, all associations stayed significant except for memory with gait speed decline. CONCLUSION: In the oldest-old, physical function and cognition are strongly related, independently of brain health. Also, the association between physical function and cognitive decline is more pronounced than the other way around, suggesting a potential for slowing cognitive decline by optimizing physical function. INTRODUCTION: Physical function and cognition seem to be interrelated, especially in the oldest-old. However, the temporal order in which they are related and the role of brain health remain uncertain. METHODS: We included 338 participants (mean age 93.1 years) from two longitudinal cohorts: the UCI 90+ Study and EMIF-AD 90+ Study. We tested the association between physical function (Short Physical Performance Battery, gait speed, and handgrip strength) at baseline with cognitive decline (MMSE, memory tests, animal fluency, Trail Making Test (TMT-) A, and digit span backward) and the association between cognition at baseline with physical decline (mean follow-up 3.3 years). We also tested whether measures for brain health (hippocampal, white matter lesion, and gray matter volume) were related to physical function and cognition and whether brain health was a common driver of the association between physical function and cognition by adding it as confounder (if applicable). RESULTS: Better performance on all physical tests at baseline was associated with less decline on MMSE, memory, and TMT-A. Conversely, fewer associations were significant, but better scores on memory, TMT-A, and digit span backward were associated with less physical decline. When adding measures for brain health as confounder, all associations stayed significant except for memory with gait speed decline. CONCLUSION: In the oldest-old, physical function and cognition are strongly related, independently of brain health. Also, the association between physical function and cognitive decline is more pronounced than the other way around, suggesting a potential for slowing cognitive decline by optimizing physical function.