To determine the accuracy of several methods for measuring the monocyte count, the results obtained by a number of different automated cell counters were analysed. Considerable discrepancies occurred for monocyte counts obtained in normal blood among the counters. The results of a visual monocyte count on a total of 800 leucocytes were used as the reference method. The technique of measuring the monocyte count by using dual staining with monoclonal antibodies CD45 and CD14 provided the closest agreement with the reference method. Six other automated counting systems were assessed. Two of these systems (Coulter VCS and Technicon H1) gave results, which, although under-estimating monocytosis, correlated well with the results obtained by the reference technique. A third system (Toa Sysmex NE-8000) gave unreliable results. Three of the automated systems evaluated measured a "third population"--that is, monocytes together with other leucocytes. One of these systems (Ortho ELT 1500), overestimated the count, as expected, but correlated well with the reference method. The second of these "third population counters" (Coulter S Plus IV) correlated moderately well with the reference monocytosis, while the Toa Sysmex E-5000 correlated poorly. It is clear that problems exist in the evaluation of different instruments for counting monocytes. An accurate and reliable reference method is a pre-requisite to evaluate this aspect of cell counters. As the visual method is too cumbersome a different reference method would be useful. Based on the results of this study, it is suggested that the technique using fluorescence labelled monoclonal antibodies should be regarded as an acceptable alternative.