Objectives: Social robots such as Paro, a therapeutic companion robot, have recently been introduced into dementia care as a means to reduce behavioural and psychological symptoms of dementia. The purpose of this study was to explore care staff perceptions of Paro and a look-alike non-robotic animal, including benefits and limitations in dementia care. Methods: The study assumed a descriptive qualitative approach, nested within a large cluster-randomised controlled trial. We interviewed a subsample of 20 facility care staff, from nine long-term care facilities in Southeast Queensland, Australia. Thematic analysis of the data, which was inductive and data-driven, was undertaken with the assistance of the qualitative software, ATLAS.ti®. Results: The findings refer to four categories: increasing excitement for Paro and decreasing enthusiasm for Plush Toy; value and function of Paro; opportunities for engagement; and alternatives vs. robustness. Conclusion: Staff caring for people with dementia preferred Paro compared to a look-alike Plush Toy. Staff identified that Paro had the potential to improve quality of life for people with dementia, whereas the Plush Toy had limitations when compared to Paro. However, participants expressed concern that the cost of Paro could reduce opportunities for use within aged care.