Four commercially available pulse oximeters were compared on normal infants to determine incidence and correct identification of movement artefact, and to assess their other features, in order to guide the potential user in the most appropriate choice of a machine to suit their purposes. Measurements of arterial oxygen saturation by pulse oximetry are distorted by movement artefact. During five one hour recording periods, over 100 artefacts occurred with each machine. The incidence of movement artefact differed between oximeters. Twice as many occurred when the Criticare 500 was used. The four oximeters differed in their ability to identify artefacts. The Novametrix 500 was best able to do this. A visual display aided observer recognition of artefacts. Other features of the different machines are described. An expansion of the use of pulse oximetry beyond the intensive care unit is suggested.