Five multichannel hematology analyzers (four different models) were evaluated to determine the ability of analyzers to correctly measure packed cell volume (PCV) values across the clinically relevant range. A sixth analyzer (fifth model) was subsequently included using archival data. Twenty-two different blood samples with PCVs ranging from 0.2 to 0.6 L/L were run in duplicate on each analyzer. Duplicate reference PCV determinations were also done using the recently described ICSH-recommended reference method. Simultaneously, a simpler "surrogate reference" procedure was performed in duplicate. Analyzer values were compared to ICSH reference and to surrogate reference PCV values using Deming regression analysis. Compared to either reference, all analyzers showed a slope of less than 1.0, thereby overestimating the PCV at 0.2 L/L and underestimating the PCV at 0.6 L/L to some extent. This outcome was surprising. Deming regression slopes would be expected to average at 1.0. Instruments whose slopes are significantly less than 1.0 will show decreased sensitivity to both anemic and polycythemic patients. We know of no inherent reason why multichannel analyzers should behave in this fashion.