The effect of abrupt medication withdrawal (no-pill discontinuation) was investigated in 1507 insomniacs using the patients' self-ratings on visual analogue scales. Drug discontinuation followed a 28-day treatment period with either 7.5 mg zopiclone, 0.25 mg triazolam, 1.0 mg flunitrazepam, or placebo in a randomized, double-blind, parallel group, multicenter study in private practice. Deterioration below individual pretreatment values (no-pill baseline) in at least one of three subjective parameters of sleep quality (sleep latency, total sleep time, nocturnal awakenings) and three parameters of daytime well-being (morning freshness, daytime tiredness, anxiety) were defined as rebound. The number of patients with rebound (rebound rate) was analyzed for every day of a 2-week posttreatment period. The overall rebound rate was higher in the placebo group (p < or = 0.001) than in each group treated with active drugs. Rebound rates affecting sleep quality were higher for placebo than for zopiclone (p < or = 0.001) and for flunitrazepam (p < or = 0.05). Rebound rates were smaller for zopiclone (p < or = 0.001) and flunitrazepam (p < or = 0.01) than for triazolam. Rebound in at least one item per day appeared in 21.5% (sleep quality) and 25.5% (daytime well-being) of the patients. Rebound decreased with increasing numbers of items of sleep quality or daytime well-being. Patients who did not respond to treatment showed higher rebound rates than those who were treatment responders (p < or = 0.001). Concerning treatment nonresponders, highest rebound was seen in the placebo group, whereas rebound was lowest in placebo responders. These results show that pill discontinuation itself may worsen sleep and daytime well-being in the sense of a rebound phenomenon. Furthermore, the number of patients with rebound remained at a high and varying level during the whole posttreatment period. This result indicates that a deterioration of sleep after drug withdrawal is not apparent during a few days but may last for longer periods in some patients and is modified by marked night-to-night variations.