Head rotation is an effective positional therapy for obstructive sleep apnea (OSA). However, not all OSA patients benefit from head rotation. This study aimed to explore the clinical phenotype of OSA patients who can benefit from head rotation. We performed a retrospective review of 184 consecutive OSA patients who underwent polysomnography. Head rotation in supine position was determined by high-quality video recording. According to the changes in apnea-hypopnea index (AHI) after head rotation, OSA patients were divided into two groups: patients with response to head rotation(HR) and patients without response to head rotation(NHR). Demographic factors and overnight polysomnography were analyzed. Compared with NHR group, HR group showed significantly lower AHI (51.8 vs 31.5, p < 0.01), time spent with oxygen saturation below 90%(5.3% vs 0.51%, p < 0.01), and higher lowest oxygen saturation(80% vs 86%, p < 0.05). Logistic regression showed that AHI was an independent factor to predict the decrease of AHI in head rotation (OR 0.985, 95% CI 0.970-0.979, p < 0.05). Among mild to moderate group (AHI < 30/h), severe group (30/h ≤ AHI < 60/h), and extremely severe group (AHI ≥ 60/h), the percentage decrease of AHI in head rotation was 18.5%, 9.5%, and 2.6%, respectively. Surprisingly, the percentage decrease of AHI of 6 responders in mild to moderate group was more than 50%. OSA patients who respond well to head rotation have less severe disease, and patients with mild to moderate OSA are more likely to improve and benefit from this position. Our research provides potential strategies and insights into the individual treatment of OSA patients.