Objectives To assess the long-term effects of the EXBELT intervention program, which was designed to reduce belt restraint use and was found to be effective immediately after implementation, after 24 months. Design Quasi-experimental longitudinal design. Setting Thirteen nursing homes: seven assigned to control group and six to intervention group. Participants A panel group (n = 225) of residents present at baseline and 24 months after baseline and a survey group (n = 689) of all residents present 24 months after baseline. The survey group included the panel group. Intervention EXBELT included four components: a policy change, an educational program, consultation, and availability of alternative interventions. Measurements The use of belt restraints 24 months after baseline was the primary outcome measure. Secondary outcomes included other types of physical restraints. An independent observer collected data four times during a 24-hour period. Results EXBELT resulted in a 65% decrease in belt use between baseline and 24 months after baseline in the panel group (odds ratio 0.35, 95% confidence interval = 0.130.93; P = .04). In the survey group, the proportion of residents using belts was 13% in the control and 3% in the intervention group (P < .001) 24 months after baseline. The use of the most restrictive types of restraints was significantly lower in the intervention group than the control group in the panel and survey groups. Conclusion The EXBELT intervention was associated with long-term minimization of belt restraint usage in older nursing home residents. A multicomponent intervention including institutional policy change, education, consultation, and the availability of alternative interventions had an enduring effect on successful restraint reduction.