Patients with sickle cell disease usually have mild hypoxaemia and their oxyhaemoglobin dissociation curve is shifted to the right. It follows that oxygen saturation in sickle cell disease should be lower than normal. Most subjects in this clinic had normal oxygen saturation by pulse oximetry, however. To improve the understanding of this paradox, arterialised capillary oxygen tension (PO2) and oxygen saturation were compared with simultaneously measured pulse oximeter saturation in 20 children with sickle cell disease. In addition, the PO2 at 50% haemoglobin saturation (P50) was compared with saturation measured by pulse oximetry in all 20 patients. It was found that saturation measured by pulse oximetry was, on the whole, similar to that calculated from the sampled blood. Individual deviations were not random, however, and were partly explained by differences in P50 values. It is concluded that pulse oximetry gives variable results in patients with sickle cell disease and should be used with caution to predict arterial saturation in this patient group.