Background: Previous empirical studies of the cognitive sequelae of failure to thrive in infancy have led to apparently inconsistent conclusions. Methods: Studies of cognitive abilities in failure to thrive were located through published bibliographies, supplemented by a search through MEDLINE. They were classified (a) into those in which the cases were identified in hospital or other specialist clinics, and those in which they were identified in primary care or by whole population screening; (b) into those that were controlled and those that were not controlled; and (c) into those with a cross-sectional and those with a longitudinal design. Effect sizes in controlled studies were summarised using D statistics for the principal cognitive outcome measure, from the last occasion on which the child was tested if the study was longitudinal. Results: In studies with cases identified in hospital or other specialist clinics (52 cases, 36 controls), the pooled effect size (weighted standardised mean difference) for cognitive outcomes was -0.85 (95% CI -0.41 to -1.30). In studies with cases identified in primary care (552 cases, 573 controls), it was -0.30 (95% CI -0.18 to -0.42). In each longitudinal study testing the same children at different ages, the effect size was smaller when the children were older. To obtain an overall estimate of the long-term cognitive outcome of failure to thrive in infancy, data from controlled studies in which cases were identified in primary care, and restricted to IQ or McCarthy scale scores in older children (502 cases, 523 controls), were used. The weighted mean difference was -0.28 (95% CI -0.16 to -0.41), equivalent to 4.2 IQ points (95% CI 2 to 6). Conclusions: Evidence from reasonably well-controlled studies indicates that failure to thrive in infancy is associated with adverse intellectual outcomes sufficiently large to be of importance at a population level.