Insulin-like growth factor-1 (IGF-1) is an essential growth factor for normal intrauterine development and postnatal growth. Mice with a complete deficiency of IGF-1 (IGF-1-null mice), created by homologous recombination, were found to exhibit postnatal lethality, growth retardation, infertility, and profound defects in the development of major organ systems. Furthermore, IGF-1-null mice were resistant to growth hormone (GH) treatment in peri-pubertal somatic growth. Using the Cre/loxP-induced conditional knockout system, we generated a mouse that lacks IGF-1 specifically in the liver, the primary site of IGF-1 production. Interestingly, although circulating and serum levels of IGF-1 were decreased by approximately 75% in these mice, they exhibited no defect in growth or development. When administered exogenously, GH stimulated IGF-1 production in several extra-hepatic tissues as well as body growth. The "Somatomedin hypothesis" originally proposed that circulating IGF-1 acting in various tissues mediate the effects of GH. These striking in vivo results, obtained using homologous recombination technology, call for a major modification of the Somatomedin hypothesis. These gene targeting studies confirm that IGF-1 is essential for GH-stimulated postnatal body growth. However, liver-derived (endocrine) IGF-1 is not essential for normal postnatal growth, though it does exert a negative feedback on GH secretion. Instead, local production of IGF-1, acting in a paracrine/autocrine fashion, appears to mediate GH-induced somatic growth. This review will discuss the effects of tissue-specific IGF-1 gene deficiency created by the Cre/loxP system versus the conventional IGF-1 knockout.