A prospective observational study was conducted in two Australian dairy herds to assess the potential for improving pregnancy rates (proportions of inseminations that result in pregnancy) to artificial insemination (AI) if the time of ovulation could be predicted with more certainty. Herd 1 calved year-round and inseminations were performed during two periods each day. Herd 2 calved during autumn'winter and inseminations were performed only after the morning milking each day. In both herds, the AI to ovulation interval of enrolled cows was determined by trans-rectal ovarian ultrasonography approximately 0, 12, 24 and 36 h after AI, and pregnancy was assessed by palpation per rectum 35'56 days after AI. Also, in Herd 1 vaginal electrical resistance (VER) measurements were taken at approximately 0, 12, 24 and 36 h after AI, and in Herd 2 cows were fitted with neck mounted activity meters that monitored cow activity count in 2-h periods. There was substantial variation in the intervals from AI to ovulation within and between herds (mean ± SD 21.2 ± 10.7, n = 102; 14.7 ± 10.4, n = 100 in herds 1 and 2, respectively). Pregnancy rates were higher for inseminations close to, but preceding, ovulation. Using combined herd data (n = 202), the highest pregnancy rate (50.8%) was observed for inseminations between 0 and 16 h before ovulation, a period in which only a modest proportion of inseminations (31.2%) occurred. In contrast, pregnancy rate was significantly lower (28.7%; risk ratio 0.6; 95% CI 0.4'1.0; p = 0.039) for inseminations between 16 and 32 h before ovulation, a period where the highest proportion of inseminations (53.2%) occurred. Thus pregnancy rates could potentially be improved if a greater proportion of inseminations were conducted shortly before ovulation. In Herd 1, mean VER during the peri-ovulatory period varied with time from ovulation. Lowest values (mean ± SEM, VER = 64.8 ± 1.2, n = 55) occurred approximately 18 h before ovulatiowere significantly lower than measurements approximately 6 h before ovulation (67.4 ± 1.0; n = 73; p = 0.003). Further work is required to determine if VER can be used to identify ovulation time and hence the optimal time to inseminate in individual animals. In Herd 2 a modest proportion of inseminations (26.9%) occurred between 24 and 40 h after the onset of increased cow activity where the highest pregnancy rate (67.9%) was observed, whereas a significantly lower pregnancy rate (42.4%; risk ratio 0.6; 95% CI 0.4'0.9; p = 0.036) was observed for inseminations between 8 and 24 h after the onset of increased cow activity where the highest proportion of inseminations (56.7%) occurred. Thus cow activity monitoring may be useful to identify the optimal time to inseminate cows. Results from this study indicate that improved methods of ovulation prediction may allow better insemination timing relative to ovulation and consequently increased pregnancy rates.