Serum prostate-specific antigen (PSA) in men is present as two different molecular forms separable by gel-filtration chromatography (GFC). We have evaluated a two-site IRMA that measures only the noncomplexed (free) form of PSA (F-PSA). Verification that the F-PSA assay measures solely F-PSA was obtained by assaying GFC-fractionated serum samples with both the F-PSA IRMA and a commercial PSA assay that measures total PSA (T-PSA: F-PSA plus alpha 1-antichymotrypsin-complexed PSA). The F-PSA assay detected only the 30-kDa peak corresponding to the free form of PSA, whereas the T-PSA assay detected two peaks: complexed PSA at approximately 90 kDa and F-PSA at approximately 30 kDa. The F-PSA assay had an analytical detection limit of 0.03 microgram/L and a measuring range up to 50 micrograms/L. The intraassay CV was 1.7-10% in the concentration range of 0.2-30 micrograms/L. The interassay CV was 3.4-12.5% in the same concentration range. Dilution and recovery studies showed no significant deviation from linearity across the assay range. The assay was insensitive to interference from hemoglobin, bilirubin, and total lipids up to concentrations of 5, 0.2, and 10 g/L, respectively. No significant loss of immunological activity (analyte stability) was seen day-to-day ( < or = 5) or after repeated freeze/thaw ( < or = 5) cycles. We conclude that the F-PSA IRMA is an accurate, precise, and reliable tool for measuring F-PSA in human serum.