The aim of the research described here was to compare different methods of body temperature (BT) measurements in dairy cows. It was hypothesised that reticular temperature (RET) values reflect the physiological status of the animals in an equivalent way to rectal (RT) and vaginal (VT) measurements. RT, VT and RET temperatures of twelve lactating Holstein–Friesian cows were measured over five consecutive days in June and October 2013. While RT and VT were manually measured three times a day, RET was automatically recorded at 10 min intervals using a bolus in the reticulum. For comparison with RT and VT, different RET values were used: single values at the respective recording times (RET-SIN), and mean (RET-MEAN) and median (RET-MED) values of 2 h prior to RT and VT measurements. Overall, body temperatures averaged 38·1 ± 0·6, 38·2 ± 0·4, 38·7 ± 0·9, 38·5 ± 0·7 and 38·7 ± 0·5 °C for RT, VT, RET-SIN, RET-MEAN and RET-MED, respectively. RT and VT were lower than all RET measurements, while RET-SIN and RET-MED were higher than RET-MEAN (P < 0·001). RET-MEAN and RET-MED values were higher in the morning, whereas RT and VT were greatest in the evening (P < 0·001). Overall, records of RT and VT were strongly correlated (r = 0·75; P < 0·001). In contrast to RET-SIN and RET-MEAN, RET-MED was higher correlated to RT and VT. In June, coefficients were higher between all methods than in October. Relation of barn T to RT and VT was stronger when compared to RET measurements. RET-SIN was higher correlated to barn T than RET-MEAN or RET-MED. Correlation between VT and barn T was strongest (r = 0·48; P < 0·001). In summary, RET-MED showed highest correlation with VT and RT. However, single RET measurements (influenced by water or feed intake) can lead to extreme variations and differences to single VT and RT values.