Background: Most developmental disorders are defined by an achievement discrepancy in which achievement on one or more specific abilities is substantially less than a person's measured intelligence. We evaluated the validity of this discrepancy criterion by assessing parameters that determine variability across abilities and by assessing relationships between achievement discrepancies and behavioral disturbances. Methods: Measures of intelligence, language, motor coordination, empathic ability, and attentional control were administered to a representative sample of 390 children aged 3 to 12 years. Parent ratings of child behavior were obtained. Results: Results indicate that achievement discrepancies are a function of the correlation between ability measures, the shape of the ability distributions, and position on an index ability dimension. Discrepancies in achievement were not related to behavioral disturbance, but underachievement relative to age peers was invariably related to behavioral disturbance. Conclusions: We conclude that developmental disorders need to be redefined in ways that are consistent with how Mental Retardation is now defined, by (a) underachievements, (b) of defined magnitude, (c) using standardized measures, (d) with known relations to normal development, and (e) concurrent deficits on standardized measures of impaired function.