Eighty-three antibodies submitted to the ISOBM TD-3 Workshop on prostate-specific antigen (PSA) were characterized by cross-inhibition studies, immunometric assay and affinity estimation with free or complexed PSA (PSA-alpha1-antichymotrypsin, PSA-ACT). Nine antibodies did not bind PSA or PSA-ACT when coated onto microtiter plates or in solution. Another 3 antibodies bound the antigens only when in solution and were therefore omitted from the cross-inhibition experiments. Dissociation constants (Kd) were estimated from the concentration of free antibody needed to achieve half-maximal binding of the antigen. Kd values for PSA and PSA-ACT ranged from 2 x 10(-12) to >10(-8) mol/l. Antibodies were classified into 6 main groups according to their reactivity. Group 1 comprised 15 antibodies (#25, 26, 33, 68, 73, 77, 78, 80, 85, 209, 213, 216, 223, 230, and 262) specific for free PSA. These antibodies had >80% cross-inhibition and showed high affinity for PSA with minimal or no affinity for the PSA-ACT complex. Group 2 comprised antibodies that reacted with both free PSA and PSA-ACT. Three subgroups were defined: group 2a (#40), group 2b (#32) and group 2c (#35, 37, 63, 90, 215 and 226). Group 3a antibodies (#31, 36, 37, 57, 64, 66, 72, 82, 84, 212, 224, 229, 257 and 260) were closely related to those of group 2, with two exceptions in group 3b (#88 and 89). Group 4 contains antibodies with binding patterns similar to those represented by groups 3b and 6b. These antibodies could be divided into two subgroups: group 4a (#30, 38, 51, 217, and 220) and group 4b (#74). Group 5 was more heterogeneous, with distinct inhibition patterns: group 5a (#50, 54, 76, 81,207, and 222); group 5b (#41), and group 5c (#28 and 86). Group 6 antibodies bind epitopes on both free PSA and PSA-ACT, but have epitopes unrelated to those represented in groups 1-3: group 6a contains 15 antibodies (#24, 27, 29, 34, 55, 56, 65, 79, 210, 214, 218, 221, 225, 258 and 261), and group 6b 2 antibodies (#67 and 75). These 6 groups represent the major immunodominant regions, one of which is exposed only on free PSA. Our classification could provide a useful guide in choosing antibodies for future PSA assays.