The socioemotional interaction with robots will become an important part of our daily lives in the future. To ensure a smooth and viable human‐robot interaction, crucial design and appearance features of robots need to be considered. However, on the one hand it is still far from clear which design criteria robots should meet to be capable of adequately expressing affective states and on the other hand if humans are capable of interpreting these affective expressions of robots correctly. Most studies on this topic use highly artificial or restricted settings. In order to examine the importance of salient robot design aspects with regard to emotional human‐robot interaction, our current experimental study investigates how different types of robots are perceived in authentic and complex settings including multimodal emotional expressions. To ensure this authenticity and complexity we used scenes from robot movies. The different scenes showed robots that systematically differed in their anthropomorphic appearance and behaviour. Participants rated these scenes regarding the robots' appearance and ability to express and convey basic emotions (fear, sadness, anger, joy, vs. neutral) in affect‐provoking situations. Results showed that participants were able to classify different types of robots portrayed in movies according to their anthropomorphic appearance, behaviour, and capability of expressing emotions. Moreover, the so‐called uncanny valley effect in robotics was found. Results are discussed in terms of their impact on future robot design and emotional human‐robot interaction. Implications for further research as well as for applied issues are outlined.