This chapter addresses the central issue of causation in epidemiology, distinguishing it from mere association. It introduces frameworks for causal inference, including the Bradford Hill criteria, which provide guidance on evaluating strength, consistency, temporality, biological gradient, plausibility, and coherence of observed associations. The counterfactual model and modern causal inference techniques, such as directed acyclic graphs (DAGs), are presented as tools to conceptualise and test causal relationships. The role of confounding, bias, and chance in distorting associations is critically examined, with emphasis on strategies to control for these factors in study design and analysis. The chapter highlights how causality is rarely established by a single study, but rather through the accumulation of evidence across multiple designs, populations, and contexts. Examples from classic epidemiological research, such as the link between smoking and lung cancer, are used to illustrate how causal arguments are built over time. By integrating historical perspectives with contemporary approaches, the chapter equips readers with both conceptual and practical tools for critically appraising evidence. This chapter maps to syllabus section 3.2.10, which requires knowledge of causation, bias, confounding, and the use of criteria such as Bradford Hill’s for causal inference.