SummaryBackgroundDecline in executive functions and related cognitive processes is associated with mobility decline, and these functions might be amenable to cognitive remediation. This study aimed to examine whether a computerised cognitive remediation programme would improve walking in adults aged 70 years and older.MethodsThis single-blind, randomised trial at one academic centre in the USA evaluated the efficacy of an 8-week computerised programme (also known as brain games) of progressive intensity and complexity to improve walking in older adults at high-risk for mobility disability. Inclusion criteria included being 70 years or older; ambulatory; and at high-risk for mobility disability, defined using a cutscore of nine or less (frail range) on the Short Physical Performance Battery and a walking speed of 100 cm/s or less. Individuals with dementia, acute or terminal medical illnesses, recent or planned surgery affecting mobility, mobility limitations solely due to musculoskeletal limitation or pain that prevented them from completing mobility tests, and those who were nursing home residents were excluded. Participants were block randomised (1:1; block size 12 and no stratification) to the intervention group or the control group (low complexity computer games and health education classes). Primary outcomes were change in walking speed at normal pace and walking while talking conditions assessed from baseline to 8 weeks post-intervention by investigators who were masked to group assignment. Groups were compared using the intention-to-treat principle with linear mixed models adjusted for confounders. This trial was registered with ClinicalTrails.gov, NCT02567227.FindingsBetween March 1, 2016, and March 12, 2020, 383 patients were enrolled and randomly assigned to the intervention or control group. After randomisation, 11 (3%) patients were diagnosed with dementia. 372 (97%; 271 [73%] women) were included in the intention-to-treat analysis. The mean age of participants was 77·0 years [SD 5·6]). 183 (49·2%) participants were Black and 62 (16·7%) were Hispanic. 314 (93%) of the target 338 completers had finished the intervention when the trial was terminated due to the COVID-19 pandemic. Although there were significant within-group improvements in both groups after the 8-week intervention, there was no significant difference in normal walking speed (−1·03 cm/s [SD 1·30]; 95% CI −3·60 to 1·54) and walking while talking conditions (0·59 cm/s [SD 1·61]; 95% CI −2·59 to 3·76) between the intervention and control groups. Similarly, within-group, but no between-group, differences were seen on executive function tests and physical function. There were no severe adverse events related to interventions.InterpretationComputerised cognitive remediation improved walking in adults aged 70 years and older at high-risk for mobility disability, but improvements were not significantly greater compared with an active control. Although our findings corroborate the within-group improvements on cognition and mobility reported in previous pilot clinical trials, future studies are required to determine the optimal dose, frequency, intensity, and content of computerised cognitive remediation programmes.FundingNational Institute on Aging.