Following Mannheim's (1970) Problem of Generations, many scholars have warned of the analytical and political risks of conflating generation with cohort. Yet the temptation persists, as relying on cohort is a convenient method of dividing a population to study it. This article proposes that cohort is only convenient if the objective is understanding generations as definitive groups of people. It suggests a supplementary objective: understanding generation as a matter of discourse. Qualitative data from interviews with 52 Canadians illustrates how the discursive forms of generation in their stories render difference, human agency and social change in atomistic or voluntaristic terms. The most extreme manifestations of this theme appear related to the perception of generational conflict. Guided by James' principle of pragmatism, this article maintains that understanding generation as a discursive, historically contingent ‘thought’ with ‘effects’ is as important as understanding its structural form and contents.