After the incubation of minced mammary tissues from non-lactating/non-pregnant (NL/NP), nonlactating/pregnant (NL/P), fully lactating (FL) and late-lactating (LL) cows with [14C]-labelled pregnenolone or progesterone and dehydroepiandrosterone (DHEA), the following metabolites were identified at all stages: 20alpha-dihydropregnenolone, progesterone (from pregnenolone), 5alpha-pregnanedione, 5alpha-pregnan-3beta-ol-20-one, 20alpha- and 20beta-dihydroprogesterone (from progesterone), 5-androstene-3beta,17beta-diol, 5alpha-androstanedione, 5alpha-androstan-3beta-ol-17-one, androstenedione, testosterone and DHEA acyl ester (from DHEA). These products indicate the occurrence of 3beta-hydroxysteroid dehydrogenase/delta5-delta4 isomerase, 17beta-hydroxysteroid oxidoreductase (17beta-HOR), 20alpha- and 20beta-hydroxysteroid dehydrogenases, steroid 5alpha-reductase and acyl transferase activities. Incubation of mammary tissue homogenates with [1,2,6,7-(3)H]androstenedione and testosterone confirmed the presence of a 17beta-HOR acting prevalently in a reductive way but failed to show evidence of any aromatase activity beyond background level. When total RNA from mammary tissues of NL/NP and LL cows was reverse-transcribed and amplified by polymerase chain reaction (PCR) with three sets of primers specific for bovine P450scc, P450c17 and P450arom cDNAs, no fragment of the expected size could be detected on gel. Southern analysis with corresponding digoxigenin-labelled ovarian probes, however, gave a positive signal for P450arom cDNA in five out of eight samples of LL mammary tissue. These data indicate that the bovine mammary gland has very limited steroidogenic capabilities that are essentially compatible with the terminal activation of circulating steroids from steroidogenic endocrines. It is uncertain, however, whether this conclusion applies to anestrous or ovariectomized lactating cows as well.