Peaks in vaginal temperature were assessed as predictors of estrus in 22 suckled beef cows during 2 breeding seasons. A flexible plastic anchor was used to retain a temperature radiotransmitter within the cow vagina and vaginal temperature was monitored at 4-min intervals. Blood was collected twice weekly for analysis of serum progesterone to confirm the occurrence of estrus and ovarian status was checked weekly using transrectal ultrasonography. Visual observations of estrous behavior were made for 20 min/h between 0400 and 0800 h and casually from 0800 to 1600 h (4 to 6 times). Values for vaginal temperature were available for 47 estrous periods. The prediction of estrus based on vaginal temperature was excellent when an estral peak in vaginal temperature was defined as an increase of at least 0.4 degree C for 3 or more consecutive hours using the corresponding hourly means of a 2 or 3-d baseline. Combining the results of 2 separate years and using a 3-d baseline, a peak in vaginal temperature was found for 42 of 47 confirmed estrus periods (detection sensitivity of 89.4% and prediction power positive of 85.7%). Corresponding detection sensitivity and prediction power positive for visual observations of standing estrus were 53.2 and 96.2%, respectively. The mean maximal increase in vaginal temperature at estrus was 0.9 +/- 0.3 degree C and the mean duration of the estrual peak in vaginal temperature was 6.5 +/- 2.7 h. In addition, vaginal temperature was found to be significantly depressed for 3 d prior to estrus and significantly elevated at mid-cycle.