Summary There is now an overwhelming preponderance of evidence that cognitive behavioural therapy for insomnia (CBT-I) is effective, as effective as sedative hypnotics during acute treatment (4–8 weeks), and is more effective in long term (following treatment). Although the efficacy of CBT-I in the treatment of chronic insomnia is well known, however there is little objective data on the effects of CBT-I on sleep architecture and sleep EEG power densities. The present study evaluated, first, subjective change in sleep quality and quantity, and secondly the modifications occurring in polysomnography and EEG power densities during sleep after 8 weeks of CBT-I. Nine free drug patients with psychophysiological insomnia, aged 33–62 years (mean age 47 ± 9.7 years), seven female and two male participated in the study. Self-report questionnaires were administered 1 week before and 1 week after CBT-I, a sleep diary was completed each day 1 week before CBT-I, during CBT-I and 1 week after CBT-I. Subjects underwent two consecutive polysomnographic nights before and after CBT-I. Spectral analysis was performed the second night following 16 h of controlled wakefulness. After CBT-I, only scales assessing insomnia were significantly decreased, stages 2, REM sleep and SWS durations were significantly increased. Slow wave activity (SWA) was increased and the SWA decay shortened, beta and sigma activity were reduced. In conclusion CBT-I improves both subjective and objective sleep quality of sleep. CBT-I may enhance sleep pressure and improve homeostatic sleep regulation.