Since 1990 the WHO Global Tuberculosis Programme (GTB) has promoted the revision of national tuberculosis programmes to strengthen the focus on directly observed treatment, short-course (DOTS) and close monitoring of treatment outcomes. GTB has encouraged in-depth evaluation of activities through a comprehensive programme review. Over the period 1990-95, WHO supported 12 such programme reviews. The criteria for selection were as follows: large population (Bangladesh, Brazil, China, Ethiopia, India, Indonesia, Mexico, and Thailand); good prospects of developing a model programme for a region (Nepal, Zimbabwe); or at advanced stage of implementation of a model programme for a region (Guinea, Peru). The estimated combined incidence of smear-positive pulmonary tuberculosis was 82 per 100,000 population, about 43% of the global incidence. The prevalence of infection with human immunodeficiency virus (HIV) was variable, being very high in Ethiopia and Zimbabwe, but negligible in Bangladesh, China, Nepal and Peru. The programme reviews were conducted by teams of 15-35 experts representing a wide range of national and external institutions. After a 2-3-month preparatory period, the conduct of the review usually lasted 2-3 weeks, including a first phase of meetings with authorities and review of documents, a second phase for field visits, and a third phase of discussion of findings and recommendations. The main lessons learned from the programme reviews were as follows: programme review is a useful tool to secure government commitment, reorient the tuberculosis control policies and replan the activities on solid grounds; the involvement of public health and academic institutions, cooperating agencies, and nongovernmental organizations secured a broad support to the new policies; programme success is linked to a centralized direction which supports a decentralized implementation through the primary health care services; monitoring and evaluation of case management functions well if it is based on the right classification of cases and quarterly reports on cohorts of patients; a comprehensive programme review should include teaching about tuberculosis in medical, nursing, and laboratory workers' schools; good quality diagnosis and treatment are the essential requirements for expanding a programme beyond the pilot testing; and control targets cannot be achieved if private and social security patients are left outside the programme scope. The methodology of comprehensive programme review should be recommended to all countries which require programme reorientation; it is also appropriate for carrying out evaluations at 4-5-year intervals in countries that are implementing the correct tuberculosis control policies.