Introduction The measurement of the net-acid-base-excretion (NABE) in urine samples of dairy herds is a tool to evaluate the risk of metabolic disorders such as milk fever. Urine specimens in large herds are causing a lot of costs and hours of labor, which can be minimized by collecting pool samples. A number of 10 single samples of cows in the same stage of lactation merged to a pool are recommended in the literature. It is well known, that the validity of pool samples is depending on the number of single samples. Based on a practical example, the aim of the study was to demonstrate these mathematical principals. In consequence of this, the diminishing significance of pool samples with a decreasing number of single samples shall be shown. Material and methods Within 4 different stages of lactation, urine samples of 10 dairy cows were collected by catheter. Each single sample was assigned to a certain pool according to a special protocol so that 10 pools of every stage of lactation and of decreasing sample size were generated and could be compared to the standard of 10 samples per pool. The evaluation of the NABE was executed with fractionated titration of the urine samples. The deviation of each pooled sample from the standard was demonstrated by comparing 95 % confidence interval (Cl). Results and conclusion The results showed that pools with declining number of samples consequently could not match the standard within the 95 % Cl. Pooled samples of 8 or less single samples proved to deliver unreliable results compared to the standard. Based on the simulation of this experimental design, the significance of NABE-values will be reduced by pooling the samples so that the test should be applied with precaution in small sized Austrian herds.