To test the effects of individual, nonfacilitated sessions with PARO (version 9), when compared against a look-alike plush toy and usual care, on the emotional and behavioral symptoms of dementia for people living in long-term care facilities. Parallel, 3-group, cluster-randomized controlled trial conducted between June 14, 2014, and May 16, 2015. Twenty-eight long-term care facilities operated by 20 care organizations located in South-East Queensland, Australia. Four hundred fifteen participants aged ≥60 years, with a documented diagnosis of dementia. Stratified by private/not-for-profit status and randomized using a computer-generated sequence, 9 facilities were randomized to the PARO group (individual, nonfacilitated, 15-minute sessions 3 times per week for 10 weeks); 10 to plush toy (same, but given PARO with robotic features disabled); and 9 to usual care. Treatment allocation was masked to assessors. Primary outcomes were changes in levels of engagement, mood states, and agitation after a 10-week intervention, assessed by coded video observations (baseline, weeks 1, 5, 10, and 15) and Cohen-Mansfield Agitation Inventory–Short Form (baseline, weeks 10 and 15). Analyses followed intention-to-treat, using repeated measures mixed effects models. Australian New Zealand Clinical Trials Registry (ACTRN12614000508673). Video data showed that participants in the PARO group were more verbally 3.61, 95% confidence interval (CI): 6.40–0.81, P = .011 and visually engaged (13.06, 95% CI: 17.05–9.06, P < .0001) than participants in plush toy. Both PARO (−3.09, 95% CI: −0.45 to −5.72, P = .022) and plush toy (−3.58, 95% CI: −1.26 to −5.91, P = .002) had significantly greater reduced neutral affect compared with usual care, whilst PARO was more effective than usual care in improving pleasure (1.12, 95% CI: 1.94–0.29, P = .008). Videos showed that PARO was more effective than usual care in improving agitation (3.33, 95% CI: 5.79–0.86, P = .008). When measured using the CMAI-SF, there was no difference between groups. Although more effective than usual care in improving mood states and agitation, PARO was only more effective than a plush toy in encouraging engagement.