For six successive milkings, six dairy cows were relocated immediately before milking to an unfamiliar operating theatre, a procedure previously shown to inhibit oxytocin release and milk ejection. Two control milkings were performed in familiar surroundings. After milk flow had ceased, two i.v. injections of 1 i.u. oxytocin were given to remove the remaining milk. Milk flow was recorded continuously and blood samples were taken every minute during milking and 10 min after milking. During the first milking in unfamiliar surroundings, no oxytocin was released. Thereby, only 13% of the total milk yield, the cisternal milk, was available and the alveolar milk fraction could only be removed after injection of oxytocin. During subsequent relocations oxytocin release steadily increased toward the control level, although the timing of oxytocin release remained delayed as compared with controls. However, the milk fraction available before oxytocin injection increased with increasing number of removals, following an asymptotic approach to control levels. The concentrations of β-endorphin, cortisol (and perhaps also of prolactin) gradually declined with the number of times the animal was moved to unfamiliar surroundings, i.e. hormone concentrations gradually adjusted to control level. During milking, concentrations of prolactin and cortisol increased, while β-endorphin concentrations decreased (except for the first relocation). We conclude that milking-related oxytocin release and therefore milk ejection adapted gradually to repeated relocations to unfamiliar surroundings. This adaptation was inversely related to β-endorphin concentrations, so it is possible that oxytocin release was suppressed by high circulating β-endorphin concentrations.