Cisternal and alveolar milk fractions were measured in East Friesian crossbred dairy ewes (n = 32) after 4, 8, 12, 16, 20, or 24 h of milk accumulation in a 6 x 6 Latin square design by administration of an oxytocin receptor antagonist for recuperation of cisternal milk followed by injection of oxytocin to remove the alveolar fraction. Less than half (38 to 47%) of the total milk yield was stored within the cistern for the first 12 h of udder filling compared with up to 57% after 24 h of udder filling. Subsequent milk yield was significantly reduced following the 16-, 20-, and 24-h treatments. Cisternal milk fat percentage, but not milk protein percentage, was lower than in alveolar milk (4.49 vs. 7.92% milk fat, respectively), indicating that casein micelles pass more freely from the alveoli to the cistern between milkings compared with fat globules. Alveolar compared to cisternal somatic cell count was higher for the 16-, 20-, and 24-h treatments. Significant increases in cisternal milk yield and milk composition observed for the 24-h compared with the 20-h treatment demonstrated the importance of the cistern as a storage space when the alveoli and small intramammary ducts became full. The main difference between cisternal and alveolar milk fractions is the poor fat content of cisternal milk, which is an important reason for the milk ejection reflex to be present during machine milking of dairy ewes. In a second experiment, milking every 16 h compared with every 12 h during mid- to late-lactation did not effect milk yield, milk composition, and quality, or lactation length; however, a 25% savings in labor was achieved with the longer milking interval.