The requirement for accuracy in the counting of somatic cells in milk has been increased by the use of this measurement as a basis of payment for milk. This paper reports on several factors that can substantially affect efficiency of staining and cell detection in various types of fluorescent cell counters. These mordant factors include sample age, chemical preservative, other sample preservation treatments and reagent properties. Experiments were conducted in several laboratories with Fossomatic 215 and 360 cell counting instruments to measure the effect of these mordant factors on both cell fluorescence and cell count. Counting efficiency was as low as 20% for some instruments counting some particular samples, depending upon the combination of adverse mordant factors. The problem of poor counting efficiency was also not detected by calibration samples that had more favourable mordant treatments applied. The adverse mordant factors were shown to reduce the intensity of fluorescence of cells for all instruments. We suggest that a significant reduction in counting efficiency due to these adverse mordant factors is most likely to occur when the particular instrument already has an intrinsic inability to separate cell fluorescence properly from background noise. This paper also discusses quality control samples, the impact of various factors on discriminator curves and the use of appropriate measures to maintain good calibration adjustment.