Obstructive sleep apnea (OSA) is a widely prevalent disorder that can affect cognitive function. The relationship between cognitive function and OSA is known to be affected by an individual's premorbid cognitive ability. Tools to measure premorbid intelligence across OSA disease severity have not been validated. This brief report aims to establish if the National Adult Reading Test (NART) provides a stable estimate of premorbid intelligence across levels of OSA disease severity. We examined if NART scores varied systematically across levels of untreated OSA severity (defined according to the apnea–hypopnea index [AHI]) and mean oxygen saturation in sleep clinic (n = 121) and community samples (n = 398) using regression analysis. Simple linear regression was used to predict NART scores based on the AHI. NART-estimated premorbid IQ scores without demographics did not vary systematically with AHI (F < 1; β = 0.01) or mean SpO2(F < 1; β = 0.12). NART-estimated premorbid IQ scores with added demographic information also did not vary systematically with AHI (F < 1; β = −0.01) or mean SpO2 (F < 1; β = 0.15). This preliminary examination shows that the NART provides a stable estimate of premorbid intelligence across untreated OSA disease severity, as demarcated by AHI or mean nocturnal SpO2.