Purpose Obstructive sleep apnoea (OSA) represents a risk for dyslipidaemia. Obstructive respiratory events during rapid eye movement (REM) sleep are more strongly related to the development of hypertension and diabetes than in non-REM. However, the relationship between sleep phases and serum lipid profile is unclear. We aimed to analyse the relationship between obstructive respiratory events in REM and non-REM sleep as well as serum lipid profile. Methods Polysomnography was performed in 94 adult subjects who did not take any lipid-modifying medications. Fasting venous blood sample was taken the following morning for total cholesterol, triglyceride, high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol, lipoprotein(a), apoprotein A1 (ApoA1) and for apoprotein B (ApoB) measurements. Lipid profiles were correlated with apnoea-hypopnoea index (AHI) during REM (AHI(REM)) and non-REM (AHI(NREM)) stages in all subjects. In addition, lipid profiles were compared between REM-dependent OSA patients (AHI(REM) >= 5/h, but AHI(NREM) < 5/h) and control subjects (both AHI(REM) and AHI(NREM) < 5/h). Results AHI(REM) correlated only with triglyceride concentrations (p = 0.04, Spearman's rho, rho = 0.21). In contrast, there was a significant association between AHI(NREM) and triglyceride (p = 0.02, rho = 0.23), ApoB (p = 0.03, rho = 0.21), HDL-C (p < 0.01, rho = -0.32) as well as ApoA1 levels (p = 0.04, rho = -0.21). However, these correlations were not present after adjustment for BMI (all p > 0.05). There was no difference in the lipid profile of REM-dependent OSA subjects and healthy controls (p > 0.05). Conclusions Altered serum lipid profile is equally associated with a disturbed REM and non-REM sleep in OSA. Obesity must be considered as a strong covariate when interpreting lipid data in sleep apnoea.