Background A large literature in developing countries finds a strong association between stunting in early childhood and educational attainment and/or cognitive performance among children of school‐going age. We contribute to the literature on the effects of stunting in childhood by exploring the links between linear growth retardation and measures of development among preschool‐aged children. Methods We analyse the association between stunting (height‐for‐age z‐score <−2) at age 2 years and children's scores on the Vineland Social Maturity Scale (VSMS) at age 4 years, a measure of social competence or ‘daily living skills’, and the Revised‐Denver Prescreening Developmental Questionnaire (R‐DPDQ) at age 5 years, a test which places greater emphasis on cognitive functioning. The sample is drawn from the Birth to Twenty cohort study, a prospective dataset of children born in 1990 in urban South Africa. We conduct multivariate regression analysis controlling for socio‐economic status, various child‐specific characteristics, home environment and caregiver inputs. Results No significant association between stunting and children's performance on the VSMS, but a large and significant association with the R‐DPDQ scores, was found. A disaggregated analysis of the various components of the scores suggests that children with low height‐for‐age at 2 years do not fall behind in terms of daily living skills or social maturity, but do substantially worse on measures capturing higher order fine motor skills and cognitive functioning. Conclusions Stunting in early childhood is strongly related to impaired cognitive functioning in children of preschool age, but does not seem to affect social maturity, at least as measured by the VSMS. These relationships between stunting at 2 years and psychosocial development at 4 and 5 years hold with extensive controls for socio‐economic status, home environment, caregiver inputs and child characteristics included in the multivariate analysis.