Objective: Fractional tissue oxygen extraction (FTOE) is a measure derived from cerebral near-infrared spectroscopy (NIRS) and simultaneous pulse oximetry (SpO2), capturing the proportion of oxygen delivered in arterial blood that is used by the target tissue. FTOE may provide a better proxy measurement of retinal hyperoxia than pulse oximetry alone and could provide insight into the risk for retinopathy of prematurity (ROP). In this study, we directly compared hyperoxia burden calculated from FTOE with hyperoxia burden calculated from SpO2 alone in order to assess the strength of association between hyperoxia and severe ROP. Study design: Infants born before <30 weeks and weighing <1500 g underwent synchronized SpO2 and FTOE recording over the first 4 days following birth. After error correction of the raw recording, hyperoxia burden was calculated as the percentage of the total SpO2 or FTOE recording with measurements exceeding defined thresholds (90/93/95% and 20/15/10%, respectively) and was compared with the outcome of severe ROP, defined as ROP requiring laser therapy, after controlling for important covariates.Result:A total of 63 infants were included with a mean±s.d. gestational age of 25.8±1.5 weeks and birth weight of 898.5±206.9 g; 13/63 (20%) had severe ROP. SpO2 hyperoxia burden was not associated with severe ROP at any threshold. FTOE hyperoxia burden was associated with severe ROP at the 15% (P=0.04) and 10% (P=0.03) thresholds. Infants with severe ROP spent 20% and 50% more time exceeding the 15% and 10% thresholds, respectively, as compared with those without severe ROP. Conclusion: In the first 96 h of life, FTOE but not SpO2 hyperoxia burden is associated with severe ROP. These preliminary results suggest that NIRS may be a viable alternative technology for targeted oxygen saturation guidelines.