The advance of artificial intelligence technologies has enabled chatbots to be emotionally responsive. While expressing empathy constitutes a critical component of emotional responsiveness often required for human service employees, the impact of empathy expressions by service chatbots is underexamined. In this research, we investigate the effect of service chatbots’ empathy expressions towards two possible sources for customers’ negative emotions: negative consumption experience and chatbot failure. Drawing on the social perception literature, we propose that chatbot-expressed empathy towards negative consumption experience enhances service evaluations by increasing perceived warmth of a chatbot, but not competence. We further propose that chatbot-expressed empathy towards chatbot failure hurts service evaluations by decreasing perceived competence of a chatbot, but not warmth. Results from laboratory experiments provide suggestive evidence for our arguments. Our theoretical framework and findings illuminate the role of empathy expressed by service chatbots and offer guidance on when to deploy empathic chatbots in practice.