In order to measure the prevalence of developmental delay among US infants and children, two types of questions were asked of parents in the 1994–95 National Health Interview Survey on Disability (NHIS-D). To measure functional delay (FD), questions from the Functional Developmental Growth Chart (FDQ), which measures specific age-appropriate tasks, were used. General delay (GD) was defined using the general type of questions about developmental delay that had been used in previous surveys. Using a nationally representative sample of 15 291 infants and children aged 4–59 months from the NHIS-D, analyses revealed that, according to these questions, ≈ 3.3% had FD and 3.4% of the children had GD. However, only one-third of the children were identified by both sets of questions. Thus, two-thirds of the children identified as having FD were not recognised by their parents as having a delay. Conversely, many parents responded to the GD questions indicating that their child had a delay, but failed to indicate that their child had a functional problem. In addition, only 17% of the children with FD and 31% of those with GD were receiving special services. Multivariable logistic regression analyses found that children with both FD and GD were more likely to be male and to be living in families with incomes below 200% of the poverty level. The findings suggest that the general types of developmental delay questions used in national surveys may not identify children with functional delays. As parents failed to identify these children, it is possible that many of these children may be slipping through paediatric surveillance. Further research to evaluate the use of these measures in population surveys is recommended.