Six chromogenic substrates for horseradish peroxidase (HRP) as an enzyme label were compared with regard to the sensitivity obtainable with a testosterone enzyme immunoassay system. The chromogens tested were 2, 2'-azinobis(3-ethylbenzothiazoline-6-sulfonic acid)(ABTS), o-phenylenediamine (OPD), 5-aminosalicylic acid (5-AS), 3-amino-9-ethylcarbazole (AEC), 3, 3', 5, 5'-tetramethylbenzidine (TMB) and 3-methyl-2-benzothiazolinone hydrazone (MBTH); for comparison, a fluorimetric assay using 3-(p-hydroxyphenyl)propionic acid was also carried out. An HRP-labeled antigen was prepared by the N-succinimidyl ester method. The bound and free enzyme-labeled antigens were separated by a double antibody method. A dose-response curve with a satisfactory sensitivity was obtained in each system by the use of a minimum amount of the HRP label at an appropriate dilution of anti-testosterone antiserum (Ka=2×1010M-1). The amounts of testosterone needed to displace 50% of the bound label ranged from 9 to 150 pg. The sensitivity decreases in the order : TMB>OPD>ABTS>5-AS>MBTH>AEC. The assay using the non-mutagenic substrate, TMB, gave a high sensitivity, comparable to that of the fluorimetric method.