Six middle aged subjects complaining of chronic insomnia associated with dysthymia were investigated in a 2-month single blind study: a 7-day placebo treatment period, followed by a 6-week phase with increasing doses of trazodone controlled release (CR) formulation (50 mg through days 8-10; 75 mg through days 11-13; 150 mg through days 14-49) and then a final 7-day withdrawal period under placebo. Medication was always administered at bedtime. Five polysomnographic recordings were accomplished by each subject (sleep 1: under baseline placebo; sleep 2-3-4; under active treatment; sleep 5: after drug discontinuation). A "blind" EEG reader analysed the traditional polysomnographic variables (macrostructure of sleep) and the amount and percentage ratio (CAP rate) of cyclic alternating pattern (CAP), the microstructural parameter that measures the instability of arousal during sleep. Visual analogue scales (VAS) for the evaluation of subjective sleep quality and the Hamilton rating scale for depression (HAM-D) were regularly assessed across the study. Statistical analysis was based on an ANOVA test with repeated measures completed by means of Bonferroni adjusted probabilities. No significant differences emerged from the macrostructural parameters referred to sleep initiation and maintenance, while significant overall modifications emerged from stage 2 (P