Background: The use of restraints in care for the elderly is sparsely described, with the exception of the use of mechanical restraints., Material and Methods: The use of restraints was investigated by interviews with the primary nurses of 1926 patients in Norwegian nursing homes. Five categories of restraint were mapped: mechanical restraints, non-mechanical restraints, electronic surveillance, the use of force or pressure in medical treatment or examination, and the use of force or pressure in activities of daily life (ADL). In addition, assumed explanatory variables were registered., Results: 45% of patients in sheltered units for the demented and 37% of those in regular wards were subjected to at least one type of restraint during a given week. The most common restraints were the use of force or pressure in ADL, the use of mechanical restraints, and the use of force or pressure in medical treatment or examination (particularly adding medicine to a patient's food or beverage). Cognitive decline, poor ADL performance, and the presence of aggressive behaviour were factors strongly associated with the use of restraints. In 65% of the cases, no written documentation could be found. In 44%, the nurse in charge made the decision to use restraint. In 20% of the cases in which medication was added to a patient's food or beverage, the decision was made by a physician., Interpretation: Restraint is frequently used in Norwegian nursing homes. High frequency, poor documentation and the arbitrariness of decision making make it reasonable to assume that the use of restraint is more often a result of inadequate routines than carefully judgment of each case.