This paper describes a radioimmunoassay (RIA) for arginine-vasopressin in which o-phenanthroline effectively inhibits cystyl-amino-peptidase activity in whole blood and plasma from pregnant women but in which o-phenanthroline is removed during the extraction of vasopressin from plasma to prevent disturbance of the RIA. Cystyl-amino-peptidase causes immediate degradation of vasopressin unless cystyl-amino-peptidase enzyme inhibitors such as o-phenanthroline are applied. However, o-phenanthroline interferes with RIA. We report an extraction procedure over octadecasyl silica-packed Sep-Pak C18 columns, by which cystyl-amino-peptidase as well as most of the cystyl-amino-peptidase inhibitor is removed from plasma with chloroform. The average o-phenanthroline concentration (0.25 mmol/l) found in the assay medium after extraction appeared not to interfere with the RIA. Polyacrylamide gel isoelectric focusing of extracts of platelet-rich and platelet-poor plasma from pregnant women revealed a single vasopressin immunoreactive peak in the RIA. Recovery and between-assay coefficients of variation of 3.2 ng/l vasopressin from pregnancy whole blood were comparable with non-pregnant controls (57%/8% and 59%/13%, respectively). Results with this assay compare well with those of another assay using inhibitors in pregnant subjects and with results in non-pregnant subjects.