Abstract Aims Investigate the relationship between global cognition and cognitive domain function and balance performance in a large sample of older people with cognitive impairment. Methods Three hundred and nine community-dwelling older people (mean age=82 years; 47% female) with cognitive impairment were recruited for the iFOCIS fall prevention randomised controlled trial. Baseline assessments completed before randomisation were used for analyses and included the Addenbrooke’s Cognitive Examination-III (ACE-III; global cognitive function) and its individual cognitive domains (attention; memory; verbal fluency; language; visuospatial ability) and executive function, further examined using the Frontal Assessment Battery (FAB). Balance performance was derived by averaging postural sway on floor and foam, maximal balance range (reverse z-score) and co-ordinated stability z-scores. With balance performance as the dependent variable, global cognition and each cognitive domain were entered into multivariate linear regression models. Results Mean (± standard deviation) ACE-III and FAB scores were 62.8±19.2 and 11.4±4.6 respectively. In linear regression analyses adjusted for covariates, global cognitive function and each cognitive domain were significantly associated with balance performance. Executive function (verbal fluency; β=-.248, p<0.001, adjusted R2=0.376) and visuospatial ability (β=-.250, p<0.001, adjusted R2=0.381) had the strongest and memory the weakest (β=-.119, p=0.018, adjusted R2=0.334) association with balance. Visuospatial ability remained significantly associated with balance performance when adjusted for attention, memory, language, verbal fluency and the FAB. Executive function (verbal fluency) remained significantly associated with balance when adjusted for attention, memory, language and visuospatial ability. Conversely, attention, memory, and language did not withstand adjustment for visuospatial ability or executive function. Conclusions Poorer global cognition and performance in each cognitive domain were associated with poorer balance performance in this large sample of community-dwelling older people with cognitive impairment. Visuospatial ability and executive function were independently associated with balance, highlighting the role higher-level cognitive processes and spatial perception and processing play in postural control.