Insulin-like growth factor (IGF)s circulate in plasma as large mol wt proteins bound to specific proteins, termed IGF-binding proteins (IGFBPs). As IGFBPs have been shown to produce artifacts in IGF radioligand assays, various extraction procedures have been proposed to eliminate IGFBPs from biological samples before radioligand assays. Comparison of acid-ethanol and C18 Sep-Pak extraction methods, the two most widely used procedures for separation of IGFs from IGFBPs in human serum samples, with the established gold standard (Sephadex G-75 acid gel filtration) revealed that a significant amount of IGFBP activity survived the acid-ethanol extraction and C18 Sep-Pak separation techniques. We, therefore, have developed a simple novel method comprising a combination of two techniques, involving separation based on size and separation based on centrifugation. In this method, serum samples were acidified and applied to Bio-Spin columns containing BSA-pretreated Bio-Gel Polyacrylamide-10 (P-10). Upon centrifugation, IGFBPs eluted in the void volume. IGFs were then eluted with 1 mol/L acetic acid containing 0.1 mol/L NaCl upon subsequent centrifugation. The efficacy of Bio-Spin P-10 separation for the complete removal of IGFBPs was determined by Western ligand blot analysis and determination of IGFBP-3 levels by RIA in the extracted serum samples. The recovery of exogenously added IGF-I to the serum samples was greater than 90%. Comparison of IGF-I and IGF-II values determined in 12 human serum samples after Bio-Spin P-10 separation with those obtained after separation with the established gold standard method (Sephadex G-75) revealed a correlation greater than 0.9. In contrast to the established gold standard method, which is tedious and time consuming, Bio-Spin P-10 separation offers the advantage of speed, such that 50 or more samples can be processed in less than 4-6 h. Application of Bio-Gel P-10 gel filtration to determine the IGF-I and IGF-II levels in 14 normal and 15 age-matched postmenopausal osteoporotic women revealed that 1) both IGF-I and IGF-II levels were reduced by 30% (P < 0.01) and 20% (P < 0.05), respectively, in osteoporotics; and 2) both IGF-I and IGF-II levels correlated with bone mineral density in the pooled data from normal and osteoporotic populations even when age was held constant (P < 0.05).