We examined the efficacy of tympanic (T(ty)) and exhaled breath (T(X)) temperatures as indices of rectal temperature (T(re)) by applying heat (condition A) and cold (condition B) in a dynamic A-B-A-B sequence. Fifteen healthy adults (8 men; 7 women; 24.9 +/- 4.6 years) volunteered. Following a 15 min baseline period, participants entered a water tank maintained at 42 degrees C water temperature and passively rested until their T(re) increased by 0.5 degrees C above baseline. Thereafter, they entered a different water tank maintained at 12 degrees C water temperature until their T(re) decreased by 0.5 degrees C below baseline. This procedure was repeated twice (i.e. A-B-A-B). T(ty) demonstrated moderate response delays to the repetitive changes in thermal balance, whereas T(X) and T(re) responded relatively fast. Both T(ty) and T(X) correlated significantly with T(re) (P < 0.05). Linear regression models were used to predict T(re) based on T(ty) and T(X). The predicted values from both models correlated significantly with T(re) (P < 0.05) and followed the changes in T(re) during the A-B-A-B thermal protocol. While some mean differences with T(re) were observed (P < 0.05), the 95% limits of agreement were acceptable for both models. It is concluded that the calculated models based on tympanic and exhaled breath temperature are valid indicators of core temperature.