Background Increasing evidence indicates that harmful effects are associated with the use of physical restraint. Objectives To characterize the use of physical restraint in intensive care units. Prevalence, adherence to protocols, and correlates of the use of physical restraint were determined. For comparisons between ICUs, adjustments were made for differences in patients' characteristics. Methods A prospective, cross-sectional, observational multicenter study with a representative sample (n = 25) of all Dutch intensive care units, ranging from local hospitals to academic centers. Each unit was visited twice, and all 379 patients admitted during these visits were included and were examined for use of physical restraint. Results Physical restraint was used in 23% of all patients (range, 0%-56% for different units). Of all 346 nurses interviewed, 31% reported using a protocol when applying physical restraint. When corrections were made for clustering within units, the risk for use of physical restraint was increased in patients with delirium or coma, in patients who could not communicate verbally, and in patients receiving psychoactive or sedative medications. Sex, severity of illness, and nurse to patient ratio were not independently related to use of physical restraint. In 11 units (44%), use of physical restraint was more frequent than expected on the basis of patients' characteristics, although this finding was not significant. Conclusions Physical restraint is frequently used in Dutch intensive care units. The differences in frequency between units suggest that opportunities exist to limit the use of physical restraint.