Although episodes of neck myoclonus (head jerks) in REM sleep have a characteristic appearance, they have so far not been described systematically in video-polysomnography. This study assesses the occurrence, frequency, and characteristics of neck myoclonus in REM sleep in a prospective sleep disorder cohort, and investigates clinical correlates and associations with medication. University hospital sleep disorders center. Two-hundred twenty-eight mixed sleep disorder patients. Not applicable. REM sleep was screened visually for short "stripe-shaped" movement-induced artifacts visible vertically over the EEG leads in polysomnographic registration. If such artifact was present, the synchronized video was inspected for the presence of neck myoclonus. Out of 205 patients, 54.6% (n = 112) had neck myoclonus during REM sleep. The mean neck myoclonus index was 1.0 +/- 2.7/h REM sleep. Younger patients had a higher neck myoclonus index than older patients (< 45 years versus 45-60 years versus > 60 years: 1.8 +/- 4.2 versus 0.6 +/- 1.1 versus 0.5 +/- 1.1; P = 0.004). Ninety-five percent of subjects < 45 years had a neck myoclonus index between 0 and 9.4/h; 95% of subjects > 45 years had a neck myoclonus index between 0 and 2.7/h. Patients on benzodiazepine treatment had no neck myoclonus (0/112 vs. 13/93; P < 0.001). In 23 patients, additional surface neck EMG was performed. EMG activation associated with neck myoclonus had a mean duration of 0.6 +/- 0.4 sec. Correlation between duration of neck EMG activation and movement-induced EEG artifact duration was very high (rho = 0.96; P < 0.001). Neck myoclonus is common during REM sleep and more frequent in younger individuals. This could indicate that neck myoclonus during REM sleep is a physiological phenomenon. If there is a cut-off distinguishing normal from excessive has to be investigated in further studies.