Linear growth from birth to 2 years of children enrolled in the World Health Organization Multicentre Growth Reference Study was similar despite substantial parental height differences among the six study sites. Within-site variability in child length attributable to parental height was estimated by repeated measures analysis of variance using generalized linear models. This approach was also used to examine relationships among selected traits (e.g. breastfeeding duration and child morbidity) and linear growth between 6 and 24 months of age. Differences in intergenerational adult heights were evaluated within sites by comparing mid-parental heights (average of the mother's and father's heights) to the children's predicted adult height. Mid-parental height consistently accounted for greater proportions of observed variability in attained child length than did either paternal or maternal height alone. The proportion of variability explained by mid-parental height ranged from 11% in Ghana to 21% in India. The average proportion of between-child variability accounted for by mid-parental height was 16% and the analogous within-child estimate was 6%. In the Norwegian and US samples, no significant differences were observed between mid-parental and children's predicted adult heights. For the other sites, predicted adult heights exceeded mid-parental heights by 6.2–7.8 cm. To the extent that adult height is predicted by height at age 2 years, these results support the expectation that significant community-wide advances in stature are attainable within one generation when care and nutrition approximate international recommendations, notwithstanding adverse conditions likely experienced by the previous generation.