ObjectiveThis study compared the values of variables measured by pulse co-oximetry (Masimo Radical 7; Masimo Europe Limited, UK) with those measured by a co-oximeter-enabled blood gas analyser (Siemens Rapid-point 500; Siemens Healthcare Limited, UK) in anaesthetized horses. Study designRetrospective study. AnimalsA total of 30 anaesthetized horses. MethodsIn total, 47 heparinized arterial blood samples were collected for blood gas analysis to determine haemoglobin concentration (tHb, g L–1) and percentage of haemoglobin saturation with oxygen (SaO2). Arterial haemoglobin saturation with oxygen was determined noninvasively by pulse co-oximetry (Masimo SpO2). Pulse co-oximetry also provided arterial haemoglobin concentration (SpHb) and arterial oxygen content (SpCaO2). Arterial oxygen content was calculated (CaO2) in 39 samples using SaO2 and the value of Hüfner’s constant used by Masimo Radical 7 (1.3 mL g–1). Data were compared using Bland–Altman analysis, correlation tests, accuracy root mean square (ARMS) statistics and total allowable error, where available. ResultsLow bias but wide limits of agreement (LoA) were found between Masimo SpO2 and SaO2 (bias = –1.4%, LoA = –4.0 to 1.3%), with an ARMS of 3%. Compared with tHb, SpHb showed low bias (6.2 g L–1) but wide LoA (–39.6 to 52.6 g L–1); its % bias (5.2%) was still within the 7% limits recommended by the Clinical Laboratory Improvement Amendments (CLIA) for humans. When comparing SpCaO2 and CaO2, the bias and LoA were –0.2 mL dL–1 and –6.7 to 6.2 mL dL–1, respectively. Conclusions and clinical relevanceMasimo SpO2 was acceptable for current ARMS standards, and SpHb measurements also met CLIA limits. The wide LoA in this study, however, suggest that the Masimo Radical 7 cannot be recommended as a substitute for direct measurements. As blood gas machines, pulse oximeters and co-oximeters use algorithms based on human haemoglobin, no true gold standard exists for horses.