Five patients with nocturnal myoclonus (periodic leg movements during sleep), mean age 59.6 years, were monitored polygraphically for fifteen successive nights. Using a double-blind drug study design with placebo at baseline, we investigated the effect of baclofen on these patients. All patients had the repetitive sleep-related abnormal movements during both the baseline nights and those on which baclofen had been administered. The number of movements varied during the four baseline nights, but the movements induced sleep fragmentation, i.e., very short electroencephalographic changes. Baclofen increased the number of movements but decreased their amplitude during non-rapid eye movement (REM) sleep and shortened the interval between movements. Its effect on sleep was dose related: as dosages increased, delta sleep progressively increased and REM sleep decreased. Sleep fragmentation resulting from muscle twitches decreased, as indicated by the diminution in alpha electroencephalographic arousals and K complexes. Baclofen dosages of 20 mg and 40 mg were the most efficacious.