Two methods for the determination of ovulation were compared to one ultrasonography performed 5 times a day. Time of ovulation by echography was 40 +/- 5.8 h (mean +/- SD) after the onset of oestrus. Preovulatory LH rise (two blood samples per day) began near the onset of oestrus but, in our conditions, this parameter could not be used to predict ovulation. The basal level of progesterone (two blood samples per day) was determined with a non-linear model, the timing when progesterone rose more than one SD (0.3 ng x mL(-1)) coincided with the timing of ovulation determined by echography (R2 = 0.98). This method was efficient and was used in a field trial to measure the consequences of the variability of the interval between Al and ovulation on litter size. The interval between Al and ovulation had an effect on litter size; litter size decreased by one piglet when this interval increased by 10h.