The links among narcissism, explicit (deliberate, controllable) self-esteem, and implicit (automatic, uncontrollable) self-esteem are unclear despite numerous attempts to illuminate these links. Some investigations suggest that narcissism reflects high explicit self-esteem that masks low implicit self-esteem, but other investigations fail to replicate this pattern. Here, we place the ‘mask’ model of narcissism in historical context and review the existing empirical evidence for this model. We then discuss three possible issues that might shed light on the inconsistent findings that have emerged from tests of the mask model. These issues include the unreliability of implicit attitude measures, narcissism's different associations with agentic versus communal self-views, and distinctions between grandiose and vulnerable narcissism subtypes. We also summarize several alternatives to the mask model of narcissism. Throughout, we offer suggestions for improving the study of narcissism and self-esteem and point to directions for future research on this topic.