To the Editor Autism is a major problem with apparently increasing prevalence. Accordingly, the association between labor induction/augmentation and autism reported by Gregory et al1 deserves further investigation. However, the results should be interpreted with great caution because of limitations in the study design. One of the most important limitations is that data were obtained from birth records in North Carolina that were previously “validated.” The cited validation involved assessment of agreement between selected variables in birth records and matched data from the same women enrolled in a cohort study.2 In the validation study, only 87% of the participants were successfully matched to the birth records (13% error rate). Many of the variables in the analysis by Gregory et al were actually never validated, including variables such as febrile, meconium, fetal distress, and placental abruption. Diabetes was one of the supposedly validated variables, but its κ was only 0.09. This is particularly important given that gestational diabetes has been associated with autism. More importantly, the variable induction/augmentation was never validated. This variable and the indications for induction/augmentation are often inaccurate in birth records.3,4