Using the Tandem-E, Axsym and LIA-mat assays, prostate specific antigen (PSA) was measured in pure PSA solutions of known concentrations of free and complexed PSA and in serum of patients with prostate cancer (n = 31), benign prostate hyperplasia (n = 32) and in healthy controls (n = 27). Measurements of pure PSA solutions showed that the AxSym assay exemplified typical properties of the skewed-response assay reacting more to free PSA than to the complexed PSA forms whereas the other two assays showed an equimolar-response. When PSA was measured in the serum of the tree groups, the AxSym test and Tandem-E gave similar PSA values whereas the LIA-mat test yielded significantly lower values. These results were not related to the amount of free PSA in the samples and proved that discordant PSA values between PSA assays were not mainly caused by the use of skewed- or equimolar assays. Despite these differences, receiver-operation characteristic analysis confirmed that the clinical validity of all three PSA tests did not differ.