Correlations between acetone in milk and milk production during the first nine weeks postpartum were calculated in ten individual cows and for a whole herd. The decrease in milk production due to ketonaemia was evaluated using a mathematical model of the KaleidaGraph integrate area program. For determination of milk components, a milk sample was obtained twice a week from one to nine weeks postpartum. Milk yield was negatively correlated with acetone concentration in milk. In non-clinical ketotic cows the calculated loss of milk yield due to elevation of milk acetone was 2-9% and in clinical ketotic cow it was as high as 26%. The correlation between milk components and acetone level varied at different stages of ketosis. Thus the stage of lactation and of ketosis should be taken into consideration when the correlation between milk parameters and milk production is evaluated. Prior to clinical ketosis, milk acetone was negatively correlated with milk yield, percentage of protein and percentage of lactose, whereas it was positively correlated with percentage of fat and urea concentration. In addition, high levels of acetone tended to be associated with high cell count. It is concluded that 1) ketotic cows do not achieve the maximum potential milk yield and 2) high levels of milk acetone are harmful to the udder. For high milk production, the concentration of acetone in milk should not exceed 0.05 mmol/l at any stage of lactation.