This study compared electroencephalogram (EEG) spectral analysis with standard visual scoring to assess the validity of clinical classification of arousals at respiratory event termination in children with obstructive sleep apnoea (OSA). METHODS: Twenty children (six M/14 F) aged 7-12 years, diagnosed with moderate to severe OSA participated in this study. Overnight polysomnography was performed, and sleep stages and arousals visually scored using clinical paediatric measures. The EEG was spectrally analysed in six 5-s epochs across respiratory events, namely two consecutive 5-s epochs pre-event onset and a 5s epoch post-event onset, 5-s before event termination, and two contiguous 5-s epochs post-event termination. EEG spectral power distribution was compared across respiratory events visually categorised as full cortical arousals, subcortical activations, or non-arousals using specialised software (Sleep Research System 5.0). RESULTS: There was no difference in power spectra between events in REM and NREM sleep and these were combined. There was a statistically significant fall from pre-arousal baseline values in delta and theta spectral power at respiratory event terminations associated with cortical arousals only. No change in power was detected at respiratory event terminations associated with subcortical activations or non-arousals. CONCLUSIONS: The lack of significant EEG spectral power changes at respiratory event terminations not associated with visually identified cortical arousals indicates undetected micro-arousals are not present. The results support the validity of clinical classifications of arousals at respiratory event termination.