Background: There is little research on the safety of the various types of restraint commonly used with individuals with intellectual disabilities who exhibit severely aggressive or self-injurious behaviour. Method: This study analysed the use of restraint with 209 individuals with intellectual disabilities over a 12-month period. Results: Planned restraint, the use of restraint as a component of a behaviour treatment programme (i.e. planned personal or planned mechanical restraint) was safer than crisis-intervention restraint (emergency personal or emergency mechanical). The overall rate of injuries during restraint was 0.46 injuries per hundred restraints. Restraint was applied 99.54% of the time without injury. Conclusion: Restraint was relatively safe and safer than reported in one other study. Planned restraint was safer than emergency restraint. The additional training and programme development associated with planned restraint may have contributed to the greater safety of planned restraint. Due to limited empirical data, restraint safety has yet to be established and this study suggests that restraint should be decreased and closely monitored.