Effects of six different milking intervals on the distribution of milk between cistern and alveoli were studied in a randomized, incomplete Latin Square experiment with four lactating Holstein cows. Cisternal and alveolar milk was measured by udder quarter at 4, 8, 12, 16, 20 and 24-h intervals with a 3-d interperiod of regular milking. Cisternal milk was evacuated using a cannula after injection of an oxytocin-receptor blocking agent, followed by an injection of oxytocin to remove the alveolar fraction. Milk samples from each fraction and quarter were collected for analysis. Cisternal and alveolar milk increased with milking interval and represented on average 30 and 70% of the milk stored in the udder, respectively. Fat content in alveolar milk remained constant during the first 16 h, increasing rapidly thereafter, reaching its maximum at 24 h (6·95%). Fat content in cisternal milk decreased with milking interval and reached its minimum at 24 h (0·96%). Total fat yield tended to increase for cisternal milk with longer milking intervals, but it increased markedly for alveolar milk, showing that fat globules did not pass freely from alveoli to cistern between milkings. Milk protein content was greater in rear quarters than in front quarters for both milk fractions. Milk protein content increased in the cisternal milk fraction and tended to increase in the alveolar milk fraction with longer milking intervals, but values did not differ between cisternal and alveolar fractions or between front and rear quarters. Total protein yield increased with milking interval in both fractions, indicating that casein micelles passed more freely than fat globules from the alveolar to the cisternal compartment. In conclusion, the short-term effects of milking intervals in milk composition were explained by the changes observed in alveolar and cisternal milk ratio.