One of the pathognomonic signs of ketosis in farm animals is an increased level of ketone bodies in blood (acetonemia). There are many methods to determine ketone bodies in the blood including those using sophisticate and expensive equipment the use of which is not always practical in a production environment. The research goal was to develop a new technique of instant diagnosis of ketone bodies level in the blood. The studies involved comparable cows through the winter-spring period. The blood biochemical study took into account ketone bodies and their fractions, beta-hydroxybutyric acid, acetone and acetoacetic acid, inorganic phosphorus, total calcium, total protein, glucose and alkaline reserve. The statistical analysis of the studied biochemical indices revealed that the blood glucose and serum alkaline reserve retained a high degree of correlation between each other and total ketone bodies both in clinically healthy cows and in the cows with subclinical ketosis. Based on the revealed regular patterns, a technique of calculating the level of ketone bodies in blood using the neural network analysis was proposed. As a result, a computer program has been developed