Abstract Background Increasing evidence links obstructive sleep apnea (OSA) to cognitive decline. Autonomic dysfunction assessed by heart rate variability is a promising early biomarker of cognitive impairment in populations without major neurocognitive disorder (MND). We aimed to determine whether nocturnal pulse rate variability (PRV) extracted from oximetry signal and OSA severity could predict MND onset among older OSA patients. Methods This study relied on data collected within the multicenter longitudinal Pays de la Loire Sleep Cohort , linked to health administrative data to identify new‐onset MND. We included patients ≥60 years with newly diagnosed OSA, and no history of MND or atrial fibrillation. Cox proportional‐hazards models were used to evaluate the association of MND with indices of PRV and OSA severity generated from sleep recordings. Results After a median follow‐up of 6.8 [4.7–9.4] years, 70 of 3283 patients (2.1%) had been diagnosed with MND. In multivariable Cox models, MND incidence was associated with age ( p < 0.0001), depression ( p = 0.013), and PRV assessed by the root mean square of the successive normal‐to‐normal (NN) beat interval differences (RMSSD; p = 0.008) and standard deviation of NN beat intervals (SDNN; p = 0.02). Patients with the highest quartile of RMSSD had a 2.3‐fold [95%CI 1.11–4.92] higher risk of being diagnosed with MND. Indices of OSA and nocturnal hypoxia severity were not associated with MND. Conclusions Within a large clinic‐based cohort of older patients with OSA, we found an association between oximetry‐based indices of PRV and the onset of MND. Nocturnal oximetry‐derived PRV indices could allow the early identification of OSA patients at higher risk of MND.