Background: Cognition has been related with gait speed in older adults; however, studies involving the oldest age group, where many have mobility disability and cognitive impairment, are few. The aim was to investigate the association between global cognitive function and gait speed in a representative sample of very old people, and whether the association was affected by dementia, and walking aid use. Method: This cross-sectional study included 1317 participants, mean age 89.4 years, and 68% women, from the Umeå85+/Gerontological Regional Database. Self-paced gait speed was measured over 2.4 m, with or without walking aids, and global cognitive function with the Mini-Mental State Examination (MMSE). The association between cognition and gait speed was analyzed using multiple linear regression and stratified according to dementia. The influence of missing gait speed values was explored using multiple imputation. An interaction analysis was performed to investigate the influence of walking aid use. Results: In comprehensively adjusted analyses, MMSE associated with gait speed (unstandardized β (β) 0.011 m/s, 95% Confidence Interval [CI] = 0.009, 0.013, p < 0.001) in the total sample. No association was found in people with dementia (β 0.003 m/s, 95%CI = 0.000, 0.006, p = 0.058), until missing gait speed values were compensated for by multiple imputation (β 0.007 m/s, 95% [CI] = 0.002, 0.011, p = 0.002). In interaction analysis the use of walking aids attenuated the association between cognition and gait speed (β − 0.019 m/s, 95%CI = − 0.024, − 0.013, p < 0.001). Conclusion:Global cognitive function appears to associate with gait speed in very old people. However, in people with dementia selection bias was indicated since unless missing gait speed values were accounted for no association was observed. Walking aid use attenuated cognitive load, which may not apply to walking in daily activities, and requires further investigation.