Companion-type social robots are expected to support elders socially and emotionally. Whilst studies show promising results, ethical concerns have also been raised. Yet, there are only a few studies that investigate ethical issues empirically. The current study investigates elders’ expectations about the companion-type robot Pleo and how much these were fulfilled after prolonged interaction through an ethical lens, thereby also targeting in differences between elders living independently and with assistance. In the study, N = 33 elders living with and without assistance in the community or in nursing homes interacted with the robot dinosaur Pleo as it suited them in their home environment for two weeks. Expectations regarding a) the robot’s capabilities, and b) the robot’s impact on elders’ lives were assessed beforehand by means of open-ended interview questions. After two weeks, elders rated the fulfillment of their individual expectations on a 7-point scale. Overall experiences as recorded interview after the interaction period were also evaluated. The results show that elders expected the robot to behave almost like a living being. Whilst, overall, participants expected some therapeutic effects, elders living with assistance anticipated to derive fun and enjoyment from the robot. Negative effects like undue responsibility, fading of enjoyment, or anger and frustration were not uncommon. The results are discussed in the light of their ethical implications.