Serum insulin-like growth factor I (IGF-I) levels predominantly reflect the hepatic effect of growth hormone (GH). Compared with serum GH levels, which reflect pulsatile GH secretion, serum IGF-I levels exhibit no major diurnal variation and thus provide a better estimate of integrated GH secretion in an individual patient. Measurement of serum IGF-I levels allows reliable identification of states of GH excess. In contrast, in a large proportion of adults with severe GH deficiency, serum IGF-I levels are within the normal range. Serum IGF-I levels increase markedly in response to GH administration and are often used as a surrogate variable for overall responsiveness to such treatment. Current data, however, suggest a poor relationship between changes in or levels of IGF-I and efficacy variables such as body composition, muscle function and well-being. The use of serum IGF-I as a guide during dose titration in the initial phase of treatment and during long-term monitoring of GH replacement therapy in adults, and its use as a safety marker or predictor of future morbidity and mortality are discussed here.