This study evaluated the arousal pattern and sleep fragmentation in the sleep microstructure of heavy snorers and obstructive sleep apnoea (OSA) patients. Fifteen snorers [Group A, (A + H I) </= 10], 15 mild OSA (Group B, A + H I > 10 </= 30] and 15 moderate to severe OSA (Group C, A + H I > 30) were studied retrospectively analysing the number, duration and type of arousals according to scoring rules concerning definition (including delta bursts) and length (from 2 to 60 s) of phasic arousal events. The number of arousals per hour of sleep related to respiratory events was higher in Groups B and C, whilst in Group A there was a number of arousals not related to apnoea or hypopnoea. Daytime sleepiness, present in all three groups and measured by a subjective evaluation, correlated with both the number and EEG type of arousal, but not with the duration. Statistical analysis indicated that arousal index related to apnoea or hypopnoea was the best variable for determining the sleepiness risk in OSA and snorers. Sleep microstructure analysis seems a good scoring method for the detection of sleep fragmentation and arousals in relation to abnormal respiratory events.