OBJECTIVE Evaluate the accuracy of this bedside method to determine hemoglobin (Hb) concentration in general surgery over a wide range of Hb values and to determine potential sources of error. METHODS Accuracy of Hb measurement using HemoCue (AB Leo Diagnostics, Helsinborg, Sweden) was assessed in 140 surgical blood samples using 7 HemoCue devices in comparison with a CO-Oximeter (IL 482, Instrumentation Laboratory, Lexington, MA). To analyze potential sources of error, packed red cells and fresh frozen plasma were reconstituted to randomized Hb levels of 2-18 g/dL. RESULTS In the surgical blood samples, the Hb concentration determined by the CO-Oximeter (HbCOOX) ranged from 5.1 to 16.7 g/dL and the Hb concentration measured by HemoCue (HbHC) from 4.7 to 16.0 g/dL. Bias (HbCOOX - HbHC) between HbCOOX and HbHC was 0.6+/-0.6 g/dL (mean +/- SD) or 5.4+/-5.0% (p < 0.001). Also in the reconstituted blood, the bias between HbCOOX and HbHC was significant (0.2+/-0.3 g/dL or 2.1+/-3.2%; p < 0.001). The microcuvette explained 68% of the variability between HbCOOX and HbHC. HemoCue thus underestimates the Hb concentration by 2-5% and exhibits a 8-10 times higher variability with only 86.4% of HbHC being within +/- 10% of HbCOOX. CONCLUSION. Although the mean bias between HbCOOX and HbHC was relatively low, Hb measurement by HemoCue exhibited a significant variability. Loading multiple microcuvettes and averaging the results may increase the accuracy of Hb measurement by HemoCue.