When using tests of infant development and intelligence in children born prematurely, the subject's age is commonly corrected for the degree of prematurity. However, there is disagreement: first, on whether this correction should ever be applied, and second, at what age to discontinue the adjustment. In a theoretical model, the difference between corrected and uncorrected scores in early infancy was massive and the difference remained clinically important until the age of 8.5 years in children who were born extremely prematurely. The clinical implications of using corrected or uncorrected scores were then evaluated in 174 very low birthweight children without severe sensorineural disabilities and with paired Bayley Mental Development Index (MDI) and Wechsler Preschool and Primary Scales of Intelligence (WPPSI) full scale scores. Failure to correct for prematurity reduced the mean MDI by 12.1 points but reduced the mean WPPSI by only 4.1 points. The disparity between individual MDI and WPPSI scores increased significantly with decreasing gestational age if uncorrected scores were used (P = 0.015) but not if scores were corrected. Using corrected scores, the MDI correctly predicted the WPPSI category in 86.1% of children (P less than 0.001) but in only 54.6% using uncorrected scores (the difference was not significant). It is suggested that a practical solution to the dilemma is to correct test scores for prematurity in the age range 2-8.5 years recognizing that only in extremely immature infants will uncorrected scores be substantially lower than corrected ones at a later age.