Insulin-like growth factor binding proteins interfere in the IGF-I and -II radioimmunoassays. In an attempt to overcome this problem, we have compared the use of truncated IGF-I, with reduced IGFBP affinity, and IGF-I as radioligands for IGF-I RIA measurements in serum separated by acid gel filtration or acid ethanol extraction followed by cryo-precipitation. With truncated IGF-I as radioligand the IGF-I measurements in acid gel filtrates and acid ethanol extracts were significantly correlated in healthy subjects (N = 42, r = 0.91, p less than 0.001) and in patients with acromegaly (N = 10, r = 0.85, p less than 0.01), GH deficiency (N = 10, r = 0.88, p less than 0.001) or Type I diabetes mellitus (N = 10, r = 0.90, p less than 0.001). In contrast, the IGF-I concentrations in acid ethanol extracts determined with IGF-I as radioligand did not correlate with those in acid gel filtrates using truncated IGF-I radioligand in patients with acromegaly (r = 0.61, NS) or GH deficiency (r = 0.46, NS). In the latter group the mean IGF-I concentrations measured in acid ethanol extracts were erroneously elevated by 112%. Low-affinity antibodies used for IGF-II RIA determinations failed to give reliable results in acid ethanol extracts from patients with Type I diabetes mellitus or GH deficiency. In conclusion, erroneously high IGF-I concentrations owing to binding of the radioligand to IGFBPs not completely removed by acid ethanol extraction can be avoided by the use of truncated IGF-I as radioligand.