Porcine and ovine follicular tissues were used to investigate, in vitro, the effect of charcoal-treated aqueous extract from ovine corpora lutea of pregnancy on aromatase activity as determined by the conversion of [³H]testosterone to oestradiol by follicular walls and measurement of ³H₂O release. Extract (500 μg protein) prepared from corpora lutea of day 112 of pregnancy but not extract (500 μg) prepared from ovine fetal cotyledonary tissue obtained at a similar time significantly decreased (P < 0.02) aromatase activity of pig follicles in the absence of FSH. These results demonstrate that a non-steroidal factor in the corpora lutea of late pregnancy directly inhibits aromatase activity. When the effects of different doses (300, 600 or 1200 μg) of luteal extract from corpora lutea of day 100 of pregnancy on aromatase activity of pig follicles were studied, the dose by treatment (presence or absence of FSH) interaction was not significant. Luteal extract dose at 300 μg did not affect aromatase activity but a significant decrease in activity occurred at 600 μg of luteal extract (600 versus 300 μg, P < 0.02). There was no further significant increase in the inhibitory effect with 1200 μg luteal extract. When the effects of 600 μg luteal extract from corpora lutea of days 15, 75 or 100 of pregnancy on aromatase activity of pig follicles were studied, a significant (P < 0.05) stage of pregnancy effect was detected, but the stage of pregnancy by treatment (presence or absence of FSH) interaction was not significant. No effect was noted with day 15 or day 75 luteal extract. In contrast, aromatase activity in the presence of day 100 luteal extract was significantly reduced compared with that of control (P < 0.01) and day 15 luteal extract (P < 0.05). A significant (P < 0.05) stage of pregnancy effect was also observed on aromatase activity of sheep follicles. Aromatase activity of sheep follicles was significantly reduced in the presence of day 100 luteal extract compared with that of control (P < 0.05) and day 15 luteal extract (P < 0.02). These data suggest that the stimulus triggering the synthesis of the aromatase inhibitor appears after mid-pregnancy. The aromatase-inhibiting activity was lost from luteal extract of corpora lutea of day 100 of pregnancy after treatment with proteolytic enzymes, demonstrating the proteic nature of the aromatase inhibitor. These experiments provide evidence for the existence in ovine corpora lutea of late pregnancy of a non-steroidal factor that reduces follicular aromatase activity. We propose the term aromatase-inhibiting factor or AIF to describe this activity.