Six Swedish Red and White dairy cows, producing 20-39 kg of 4% fat-corrected milk were given a ration balanced in energy and protein. They had access to feed from 05.15 to 09.00 and from 13.00 to 16.30 and were milked at 06.15 and 15.30. The milk was analysed for urea with a FIA technique. There was a significant diurnal variation in milk urea. The highest values were found 3-5 h after the beginning of the morning feeding and the lowest values (down to 60% of the max. values) during late night. Within 1 h after the start of the morning feeding the urea values had increased significantly, but they had decreased within the same time after the start of the afternoon feeding. Since there was a pronounced diurnal variation in the milk fat content, the urea concentration was also recalculated to concentration in the water phase of the milk. It was higher in that phase, but the pattern of the diurnal variation was not changed significantly. However, analyses on milk with a very high fat content may give misleading results. There were no important differences in the milk urea concentration of different udder quarters. When calculated as concentration in the water phase of the milk, no differences in urea concentration were found between the beginning and the end of milking. The analytical method had a good precision (coefficient of variation max. 3%). The milk urea concentration was not changed significantly after storage during 10 days at 4 degrees C when no preservative was added; but after 17 days the milk had turned sour and the urea value had increased. When a preservative (bronopole) was added the urea concentration remained unchanged during 17 days. Deepfreezing did not influence the urea concentration.