The pre-analytical error potential of air bubbles contaminating blood gas samples has been well recognized for blood gas tension and pH measurements [1], and is thus considered in present recommendations for blood gas sample handling [2]. The effect of air contamination on the oxygen saturation of haemoglobin (HbO), measured by co-oximetry, has however, scarcely been appreciated [1]. With increased reliance on directly measured HbO for the evaluation of blood oxygen content [3,4], a closer look at possible errors of this parameter is warranted. This study was undertaken to estimate the early effect of graded air contamination in conditions simulating pre-analytical sample handling in clinical practice.