By studying communicative interactions between humans, we can investigate the basic processes underlying the evolution of language, including how humans manage to communicate in the first place, how they form novel conventions, how they create grammatical structure, and subsequent changes to their conventions and grammar. Communication games, which involve interactions between two or more participants in artificial settings (in the lab or online), are a useful method to address these questions within a controlled experimental environment. These experiments can help researchers with teasing apart the effects of different variables (e.g., population size, context, feedback, modality, interaction history, and more) on the emergence of language, which are typically confounded in naturalistic settings.In this chapter, we first briefly review the history of the communication game paradigm, from classic referential games in experimental pragmatics (Lewis game, tangrams, maze game) to experimental semiotics. We then summarize the general principles, procedures, designs, and typical measures that characterize communication games, and introduce some of the most prominent usages of these games for language evolution research. We illustrate these points via two recent example studies showcasing the power of the communication game paradigm: one study focusing on the impact of shared context on the formation of conventions (i.e., does access to a shared visual environment promote more successful emergent communication?); and a second study testing the impact of population size on the emergence of grammar (i.e., do big and small groups create languages with more or less compositional structure?). Finally, we discuss the theoretical limitations and methodological challenges of using such paradigms, and propose some ways forward.