The time constraints of the classic twice-daily milking routine are less easily endured by individual dairy farmers, because of their impact on quality of life. Our aim was to evaluate milk production responses by dairy cows milked twice daily at contrasting intervals. In experiments 1 (20 cows) and 2 (28 cows), four milking regimes were compared during a 3-week period beginning after the peak of lactation. Three groups of five cows were milked twice daily (TDM) with milking intervals of 11 : 13, 7 : 17 and 3 : 21 h in experiment 1, and three groups of seven cows at 11 : 13, 5 : 19 and 2.5 : 21.5 h in experiment 2. One group (five and seven cows respectively) was milked once daily (ODM) in each experiment. In experiment 3 (three groups, 12 cows per group), one group was milked at 10 : 14 h and one at 5 : 19 h, and the third group once daily. Milking treatments began during the second week of lactation and continued for an average of 23 weeks. In experiments 1 and 2, daily milk yields were reduced by 4.1%, 11.5% and 28%, for the 5 : 19, 3 : 21 and ODM milking treatments compared with the 11 : 13 h interval. In experiment 3, the decrease in daily milk yields for 5 : 19 h and ODM was 10% and 40% compared with the 10 : 14 h time interval. In the average daily milk, fat and protein contents and somatic cell counts were not different between the TDM groups, and the ODM group had (or tended to have) a higher fat and protein content. For a given milking, milk fat content decreased from about 60 to 32 g/kg as the preceding milking interval increased from 2.5 to 3 h up to 12 h. It then levelled out and even increased, mainly after 18 to 20 h. Somatic cell count showed a similar trend, and protein content did not change steadily. Dry matter intake, body weight and body condition score were not affected by contrasting milking intervals. After resumption of TDM with conventional intervals, productions of milk, fat and protein no longer differed between the TDM groups. Milk yield of previously ODM cows remained lower by 2 kg/day ( P 5 0.15) in experiments 1 and 2, and by 7 kg/day ( P , 0.05) in experiment 3. These results suggest that TDM at contrasting intervals up to 5 : 19 h is feasible as it decreases milk yield only moderately, especially if implemented from peak of lactation.