Measurements of haemoglobin concentration (Hb) and haematocrit (Hct) are often used to estimate changes in blood and plasma volume, most commonly employing the calculations described by Dill and Costill (1974). Since red cell volume is highly responsive to changes in extracellular osmolality, the measurement of Hct using automated analysers that dilute the sample in a medium of fixed osmolality may not be valid when repeat measures of Hct are made after interventions that change osmolality. This investigation quantified the effect of changes in plasma osmolality on the measurement of Hct by centrifugation and using an automated analyser. With ethics committee approval, eight healthy male volunteers visited the laboratory after an overnight fast. After 15 minutes of seated rest, a 20mL blood sample was collected from a superficial antecubital vein by venepuncture. Whole blood was treated with K2EDTA (1.5mg/mL) before being aliquoted into collection tubes containing 100µL of saline of varying concentrations to alter the sample osmolality. Samples were thoroughly mixed and analysed for haemoglobin concentration using the cyanmethaemoglobin method, and haematocrit, by spun packed cell volume (PCV). Hb and Hct were also determined using an automated haematology analyser (Beckman-Coulter AC.T 5diff). Plasma osmolality was measured using freezing point depression. Changes in blood, plasma and red cell volumes were calculated as outlined by Dill and Costill (1974). The addition of saline to the samples resulted in a plasma osmolality range between 263 ± 4 and 335 ± 3mosmol/kg. Plasma osmolality did not influence Hb measurements made using the cyanmethaemoglobin method (P = 0.239) or the haematology analyser (P = 0.306). While Hct values from the haematology analyser were not different (P = 0.652), PCV was progressively lower as the osmolality of the sample increased (P < 0.001). When using the PCV data, plasma osmolality did not influence calculated changes in blood volume (P = 0.299), but a significant increase in plasma volume was apparent when osmolality increased (P < 0.001). Regression analysis revealed that a 10 mosmol/kg change in plasma osmolality would result in an apparent 1% change in plasma volume. With the autoanalyser data, however, it appeared that no change in red cell (P = 0.525) or plasma volume (P = 0.291) occurred. Conditions or interventions which result in a marked change to plasma osmolality may therefore produce a discrepancy in Hct measured using an automated analyser. This may invalidate the use of these data to determine changes in plasma volume.