The milk and body temperatures of 15 cows were monitored twice daily at milking times, over a total of 30 oestrous periods. The best determination of oestrus was based on a temperature rise of at least 0.2 degree C over the corresponding mean temperature of the three preceding days. This resulted in 72 +/- 16 per cent of oestruses, predicted by serial progesterone assays, being successfully detected by serial progesterone assays, being successfully detected and 11 +/- 3 per cent of false positives. The variable extent of the temperature rise at oestrus and the large daily fluctuations in temperature for individual cows, resulted in a moderate oestrus detection rate accompanied by an unacceptable level of false positives.