Tuberculosis (TB) is one of the leading causes of morbidity and mortality in KwaZulu-Natal (KZN), South Africa. Directly observed therapy (DOT) is a key element of the WHO's Directly Observed Treatment, Short Course (DOTS) strategy to control TB. Since the inception of DOTS in South Africa in 1996, its impact has never been assessed. We evaluated the DOT programme in the priority facilities of the four TB crises districts (EThekwini, UMgungundlovu, UMzinyathi and UThungulu) in the province of KwaZulu-Natal. A semi-structured questionnaire was used to interview TB nurses and community DOT supporters. The primary outcome used was cure rate. On average, priority facilities in districts that have high DOT coverage had better cure rates compared with those that have low DOT coverage (beta=0.818, 95% CI 0.023-1.614; P=0.045). The fewer the number of patients allocated to a DOT supporter the higher the cure rates (beta=-1.984; 95% CI -3.88 to 0.086; P=0.041). There was no difference in cure rates between facilities with tangible reporting and recording methods and those with none. These findings suggest that cure rates can be improved if DOT is implemented appropriately.