Thirty cyclic, non-suckled Brahman cows were divided into three groups, all of which were synchronized sequentially with CIDR-B and observed continuously for 100 h to determine different behavioural oestrus signs. Twenty-four hours after implant withdrawal, all synchronized cows in the group, together with all other cows displaying oestrus, were subjected to intensive ultrasonographic observations (every 6 h for 120 h) to pinpoint the moment of ovulation. In the first group, oestrus and ovulation response was 60% (6/10), in the second 44% (4/9) showed oestrus and six ovulated, and in the third group oestrus and ovulation were 80% (8/10). Significant differences were observed between the second and third groups (p < 0.05). No differences were observed in the duration of oestrus, time when oestrus was displayed after implant withdrawal, time of ovulation and onset of oestrus, end of oestrus to ovulation, and intensity of oestrus on a point scale. The relationship between duration of oestrus and time of ovulation was r(2) = 0.16. Ovulation, on average, was 32.1 +/- 14.5 h after the onset of oestrus, 22.3 +/- 16.5 h after the end of oestrus, and 91.8 +/- 16.7 after implant withdrawal, although no significant differences were observed. One non-synchronized animal showed oestrous activity in the second group but failed to ovulate. In the third group, 8 animals showed oestrus, 4 with high concentrations of progesterone. Of the other four one ovulated. In conclusion, oestrous behaviour is not necessarily the best marker to predict the time when ovulation takes place due to variation in the length of the oestrous period and the possible integration of non-ovulatory animals into sexually active groups.