A protein-binding assay for insulin-like growth factor II (IGF II) is described. The assay uses IGF binding proteins extracted from human cerebrospinal fluid which have selective affinity for IGF II. IGF I was 9 times less potent than IGF II in displacing [125I]IGF II, and when mixtures of the IGFs were assayed at IGF I/IGF II ratios of 2, 5, and 10, interference from IGF I in the assay was 0%, 5%, and 9%, respectively. Given the serum concentrations of IGF I and IGF II estimated by RIA and by this protein-binding assay, IGF I can be said to have had no cross-reaction when IGF II was assayed in human serum and at most 5% cross-reaction in the case of rat serum. After separation of IGFs from their binding proteins by acidic gel filtration, serum IGF II levels (mean +/- SE) measured by this method were 1322 +/- 66 ng/ml in normal adults, 500 +/- 65 ng/ml in patients with total GH deficiency, 1327 +/- 69 ng/ml in untreated acromegalic patients, and 1817 +/- 145 ng/ml in uremic patients undergoing chronic hemodialysis. In postpubertal young rats, the mean serum IGF II level was 43 +/- 2.6 ng/ml and after hypophysectomy it was 16 +/- 2.4 ng/ml. Although the IGF II levels in man and in the rat were different, they appeared to be similarly GH dependent, although less so than IGF I. In view of the sensitivity (0.03 ng IGF II) and the specificity of this assay, the small quantities of cerebrospinal fluid required (1 mu leq/assay tube) and its applicability for IGF II measurement in several species, the use of this assay for measuring IGF II in a variety of biological media can be envisaged.