An immunoradiometric assay (IRMA) for the measurement of insulin-like growth factor-II (IGF-II) in human plasma has been developed, optimized and evaluated clinically in normal subjects and patients with disorders of the GH/IGF-I axis. Six monoclonal antibodies (MAbs) to recombinant human IGF-II (rhIGF-II) were produced, all of which had low cross-reactivity with rhIGF-I (< 0.01%) and insulin (< 0.01%). Compatibility of pairs of MAbs was tested in two-site IRMAs using three radioiodinated MAbs and three MAbs linked to Sephacryl S-300 (with separation of bound and free radiolabelled MAb by sucrose layering). Seven pairs of MAbs bound rhIGF-II and the combination of 125I-labelled W3D9 and W2H1 linked to solid phase was selected. The optimized assay had a completion time of 4 h, a minimum detection limit of 30 ng/ml (2.5 standard deviations from the zero standard) and detected a single peak of endogenous IGF-II in normal plasma which co-eluted with rhIGF-II after acid gel chromatography. IGF-II was measured in formic acid/acetone extracts of plasma from 16 normal subjects (mean 685, range 516-1008 micrograms/l), four acromegalic patients (mean 637, range 553-700 micrograms/l), fourteen patients with type-1 diabetes (mean 635, range 247-753 micrograms/l), nine patients with uraemia (mean 423, range 78-850 micrograms/l), and three patients with Laron-type GH insensitivity (75, 35 and 36 micrograms/l). No significant fluctuations were detected between samples obtained hourly from 08.00 to 19.00 h in normal subjects. Low levels of IGF-binding proteins (IGFBPs) remaining in plasma extracts may interfere with the measurement of IGF-II and give rise to falsely elevated IGF-II levels in radioimmunoassays or falsely suppressed levels in IRMAs. Such interference did not occur with the IRMA when used to measure IGF-II in extracts from normal subjects, acromegalic patients and patients with type-1 diabetes, and the addition of excess rhIGF-I in order to displace IGF-II from residual IGFBPs had no effect on IGF-II measurements in these samples. However, levels of IGF-II measured in extracts from patients with Laron-type GH insensitivity and patients with uraemia increased markedly after preincubation with excess rhIGF-I. The accurate measurement of IGF-II by IRMA in extracts from these subjects therefore requires the displacement of IGF-II from IGFBPs prior to assay. We conclude that, in contrast to radioimmunoassays, the two-site IRMA developed here provides a practical, rapid and specific method for the measurement of IGF-II in human plasma.