We aimed to estimate the relationship between height (a measure of early-life cumulative net nutrition) and later-life cognitive function among older rural South African adults, and whether education modified this relationship. Data were from baseline in-person interviews with 5059 adults >/= 40 years in the population-based "Health and Aging in Africa: A Longitudinal Study of an INDEPTH Community in South Africa" (HAALSI) study in Agincourt sub-district, South Africa, in 2015. Linear regression was used to estimate the relationship between height quintile and latent cognitive function z-score (representing episodic memory, time orientation, and numeracy), with adjustment for life course covariates and a height-by-education interaction. Mean (SD) height was 162.7 (8.9) cm. Nearly half the sample had no formal education (46%; 2307/5059). Mean age- and sex-adjusted cognitive z-scores increased from - 0.68 (95% CI: - 0.76 to - 0.61) in those with no education in the shortest height quintile to 0.62 (95% CI: 0.52-0.71) in those with at least 8 years of education in the tallest height quintile. There was a linear height disparity in cognitive z-scores for those with no formal education (adjusted beta = 0.10; 95% CI: 0.08-0.13 per height quintile), but no height disparity in cognitive z-scores in those with any level of education. Short stature is associated with poor cognitive function and may be a risk factor for cognitive impairment among older adults living in rural South Africa. The height disparity in cognitive function was negated for older adults who had any level of education.