STUDY OBJECTIVES: Although recent studies suggest that OSA during rapid eye movement (REM) is associated with different cardiometabolic and neurocognitive risks compared with non-REM (NREM) sleep, there is no information on whether OSA during REM and/or NREM sleep is independently associated with diabetic kidney disease (DKD). METHODS: In this cross-sectional study, 303 patients with type 2 diabetes who were followed up at our diabetes outpatient clinic underwent all-night polysomnography. Logistic regression analysis was performed to determine the separate effects of OSA during REM and/or NREM sleep (REM and/or NREM-AHI), apnea-hypopnea index (AHI) and several other polysomnography parameters on DKD, after adjustment for several known risk factors for DKD. RESULTS: The median (interquartile range) AHI, REM-AHI and NREM-AHI of the patients (age 57.8 +/- 11.8 years, male sex 86.8%, hypertension 64.3% and DKD 35.2%) were 29.8 (18.0-45.4), 35.4 (21.1-53.3) and 29.1 (16.3-45.4) events/h respectively. REM-AHI quartiles, but not NREM-AHI quartiles, correlated independently and significantly with DKD (p=0.03 for linear trend, odds ratio (OR) and 95% confidence interval (CI) for Q2: 3.14 (1.10-8.98), Q3: 3.83 (1.26-11.60), Q4: 4.97 (1.60-15.46), compared with Q1). In addition, categorical AHI (p=0.01, OR and 95%CI for ≥15 to <30: 1.54 (0.64-3.71), ≥30: 3.08 (1.36-6.94) compared with <15), quartiles of AHI (p=0.01), quartiles of lowest arterial oxyhemoglobin saturation (SaO₂) (p<0.01), quartiles of percentage of time spent with SaO₂ <90 (T90SPT) (p<0.01) and quartiles of mean SaO₂ were independently associated with DKD. CONCLUSIONS: OSA, especially during REM sleep, is a potential risk factor for DKD.