Differential cell count of milk is a traditional parameter for the evaluation of udder health. The literature shows great variation in differential cell counts of the milk of healthy mammary glands: macrophages range from 0% to 80%, lymphocytes from 1·5% to 79·5%, polymorphonuclear neutrophils from 3% to 95%, and epithelial cells from 1% to 19%. We conducted three studies to seek explanations for such variation. In the first, we evaluated the impact of polyethylene and glass sampling bottles. The aim of the second study was to compare the results of differential cell counts performed by three different technicians. The third study evaluated two methods of smear preparation. When polyethylene plastic bottles were used, the macrophage population was minimized but lymphocytes remained unaffected. This was shown by an exemplary flow cytometric analysis using four monoclonal antibodies against three lymphocyte surface structures. There were significant differences in the differential cell counts of 40 smears made by three technicians despite identical operating procedures. For the sediment smear, milk was centrifuged once and the sediment spread by eye on a glass slide. For the “coffee grinder” smear method, the sample was subjected to four centrifugations and then placed on a cover glass in order to spread the sediment using centrifugal force. The coffee grinder procedure led to a reduction of lymphocytes and an enrichment of polymorphonuclear neutrophils without affecting the macrophage population. Both methods made it possible to distinguish different udder health classes. It can be concluded that differential cell counts are a useful tool for comparing and monitoring udder health only if: samples are taken in a glass bottle; smears are prepared with the identical technique; and the differential cell counts are performed by a single person.