Direct observation of patients taking their medication is a strategy to improve completion rates for tuberculosis treatment, but the programmes to implement this approach consist of a complex array of inputs aimed at influencing adherence. Policy makers need a clear understanding of these inputs to succeed. We systematically identified and reviewed published reports of direct observation therapy (DOT) programmes and compared inputs with WHO's short-course DOT programme. DOT programmes frequently consist of more than the five elements of WHO's strategy, including incentives, tracing of defaulters, legal sanctions, patient-centred approaches, staff motivation, supervision, and additional external funds. Focusing on direct observation as a key factor in the promotion of adherence seems inappropriate. Multiple components might account for the success of DOT programmes, and WHO should make these explicit.