BACKGROUND: Spurious hypoxemia has been described in case reports during extreme hyperleukocytosis and has led to recommendations for immediate cooling and analysis of arterial blood gases (ABGs). We sought to determine, in samples processed as recommended, the magnitude of spurious hypoxemia in acute leukemia subjects with hyperleukocytosis. METHODS: A retrospective chart review was conducted of all subjects admitted between 2003 and July 2014 for acute leukemia, who presented with white blood cell (WBC) count > 50 × 109 cells/L and had ABGs performed. For each ABG, we collected PaO2, SaO2, simultaneous WBC count, and SpO2 when available. Bland and Altman analysis was used to assess the agreement between SpO2 and SaO2. RESULTS: One-hundred forty-six samples (from 45 subjects) were included, of which 57 samples (from 18 subjects) had data available for Bland and Altman analysis. Mean (SpO2 − SaO2) was 2.5%, and 95% CI for limits of agreement between SpO2 and SaO2 was (−10.1,15.1)%. The mean (SpO2 – SaO2) was significantly higher for WBC count > 100 × 109/L as compared with WBC count 150 × 109/L had PaO2 94%, the proportion being 5 of 62 samples for WBC count 150 × 109/L exhibited large SpO2 to SaO2 differences (10–20%) before leukapheresis, which decreased to below 5% afterward. CONCLUSIONS: In subjects with acute leukemia and hyperleukocytosis, despite cooling and quickly analyzing the samples, we observed poor correlation and agreement between SpO2 and SaO2, unacceptably low for WBC count > 100 × 109/L. Our results suggest that current guidelines may not totally prevent the diagnosis of spurious hypoxemia.