This study sought to assess the effect of giardiasis on growth of young children. In Salvador, northeast Brazil, 597 children initially aged 6 to 45 months were followed for a year in 1998/9, measured anthropometrically thrice, every 6 months, and monitored for diarrhoea prevalence twice weekly. Stool samples were collected and examined during the second round of anthropometry, and infected children were treated 39 days later, on average (S.D. 20 days). For each 6-month interval, the gains in z-scores of infected and uninfected children were compared, after adjustment for potential confounding factors, including longitudinal prevalence of diarrhoea. No significant difference was found for the first interval but in the second, the gain in adjusted height-for-age z-score was 0.09 less in infected than uninfected children, equivalent to a difference in height gain of 0.5 cm. The shortfall in growth was greater in children who remained free of diarrhoea, and was significantly correlated with the proportion of the second interval during which the child had remained untreated. We conclude that Giardia can impede child growth even when asymptomatic, presumably through malabsorption. This finding challenges the view that young children found to have asymptomatic giardiasis in developing countries should not be treated.