Human perceptions of the “humanness” of robots have been found to be influenced by face, voice and interactivity features. These features have been studied individually in a human robot interaction (HRI) and facial features appear to be strongest in promoting positive human emotions. The objective of this study was to assess the effects of combined humanoid robot features on human emotions during a medicine delivery task. Seven robot prototypes with various combinations of face, voice and interactivity features were developed and classified in terms of levels of humanness. A “Wizard of Oz” experiment was conducted in which 32 subjects received and accepted a simulated bag of medicine from each of the robot prototypes. Both subjective (arousal and valence ratings) and physiological (HR and GSR) measures were collected as indicators of participant emotional states. Results revealed robot configurations with higher levels of humanness promoted positive emotions. Arousal and valence ratings and the HR response had utility for predicting emotions. We also found that additional humanoid features lead to higher GSR ratings, but the trend was not strictly linear with the pre-defined level of robot humanness.