Epidemiology has a variety of definitions but is distinct from other medical research in being interested in public health as well as individual health and being more concerned with causation. Psychiatric epidemiology has developed alongside advances in non-infectious disease epidemiology. Good psychiatric epidemiology should be similar to good clinical psychiatry and requires a biopsychosocial approach. To infer causation, the investigator needs to consider alternative explanations for an observed association. The role of chance, confounding, bias and reverse causation are briefly described with some examples from the psychiatric literature.