We evaluated the analytical performance of two commercial RIA kits for the assay of plasma atrial natriuretic peptide (ANP), which use different antisera and tracers, in order to verify the most suitable RIA procedure for the assay of ANP in plasma samples. In particular, two different procedures for the purification of plasma were evaluated. The first one uses the extraction of plasma samples with Sep-Pack C18 cartridges and the second immunoextraction (C-terminal ANP-specific antibody bound to solid phase of Sepharose). The sensitivity and precision of the two RIA kits proved inadequate, being unable to provide an acceptably precise measurement of the plasma ANP concentrations in normal subjects. An improvement in precision and sensitivity was obtained by using "fresh" (or purified) preparations of the tracer and standard solutions, and by adding PEG to the B/F separation step. The direct assay (without preliminary purification) was not possible due to a serious overestimation of plasma levels due to the presence of interferences; on the other hand, the extraction step increased the imprecision and the complexity of the assay. Moreover, the extraction recovery of ANP added to plasma samples in the procedure using Sep-Pak cartridges is about 50-60%, while it was found to be almost complete (about 90%) in the immunoextraction procedure. Moreover, a comparison of the results obtained with the two RIA systems indicated that the antisera have comparable sensitivities, but quite different specificities; however, neither of the two RIA kits showed a completely satisfactory degree of sensitivity, precision and practicability.