Serum levels of insulin-like growth factor-I (IGF-I) increase with age and pubertal development. The large variation in circulating IGF-I levels in adolescence makes it difficult to use the IGF-I value of a single child in the assessment of his growth status. In addition, the interference of IGF-binding proteins in many IGF-I assays contributes to this problem. We measured IGF-I in acid-ethanol-extracted serum from 1030 healthy children, adolescents, and adults, employing a RIA that reduces interference of IGF-binding proteins by using monoiodinated Tyr31-[125I]des-(1-3)IGF-I as radioligand. Mean serum IGF-I concentrations increased slowly in prepubertal children from 80-200 micrograms/L with a further steep increase during puberty to approximately 500 micrograms/L. After puberty, a subsequent continuous fall in circulating IGF-I levels was apparent throughout adulthood to a mean of 100 micrograms/L at the age of 80 yr (P < 0.0001). Girls had maximal IGF-I levels at 14.5 yr of age, whereas boys had peak IGF-I levels 1 yr later. This is almost 2 yr later than average peak height velocity. The large variation in serum IGF-I levels during puberty was diminished when data were separated according to sex and Tanner stage of puberty. Interestingly, we found a significant variation with age within the Tanner stages; there was an increase in serum IGF-I concentrations with age in the early pubertal stages and a decrease in the late stages (P < 0.05). Serum IGF-I increased concomitantly with increasing testicular volume. Multiple regression analysis revealed that serum IGF-I levels predicted height velocity in the following year (r = 0.33; P < 0.0001). Body mass index did not correlate significantly with serum IGF-I in prepubertal children in a multiple regression analysis. In conclusion, there was a significant variation in serum IGF-I levels with age within a given Tanner stage of puberty in addition to the well known increase with increasing age or pubertal stage. Accordingly, the effects of sex, age, and puberty on serum IGF-I cannot be separated into simple additive components when studying 1030 children in a cross-sectional design. Thus, the age-, sex-, and puberty-corrected IGF-I values may, in fact, improve the use of serum IGF-I as a diagnostic tool to distinguish between a child with retarded puberty and a GH-deficient individual.