In a clinic population of 289 children (nine years of age and under) suspected of developmental delay, it was found that correlations on repeated administrations of the Bayley and/or Stanford-Binet intelligence tests were much higher than in normal children; that the magnitude of the correlations was inversely related to IQ level; that children initially tested under two years of age displayed more IQ change than those tested over the age of two, with girls more likely than boys to show a drop, and the greatest amount of fluctuation occurring the 51-80 range. The IQ constancy of this population has important developmental implications for the retarded: it suggests that they are growing intellectually at a rate proportional to normal children of their own mental rather than chronological age; that they may well have more years of maturation before reaching a plateau; and that educational programs should take into consideration this lengthened learning period.