Collective emotions are at the heart of any society and become evident in gatherings, crowds, or responses to widely salient events. However, they remain poorly understood and conceptualized in scientific terms. Here, we provide first steps towards a theory of collective emotions. We first review accounts of the social and cultural embeddedness of emotion that contribute to understanding collective emotions from three broad perspectives: face-to-face encounters, culture and shared knowledge, and identification with a social collective. In discussing their strengths and shortcomings and highlighting areas of conceptual overlap, we translate these views into a number of bottom–up mechanisms that explain collective emotion elicitation on the levels of social cognition, expressive behavior, and social practices.