SETTING: Five districts in Malawi. OBJECTIVE: A new oral anti-tuberculosis treatment regimen with different directly observed treatment (DOT) choices in the initial phase of treatment was introduced for new patients in the five districts. The objectives were to determine 1) the site of DOT during the initial phase of treatment, and 2) the effectiveness of the new regimen. DESIGN: Prospective data collection on all tuberculosis (TB) patients registered in a phased approach between 1 July 1997 and 31 December 1998, including site of DOT option in initial phase of treatment, 2-month and 8-month treatment outcomes, 2-month sputum smear conversion in smear-positive pulmonary tuberculosis (PTB) patients and in-patient hospital bed days. RESULTS: There were 6335 new patients: 2671 (42%) with smear-positive PTB, 2211 (35%) with smear-negative PTB and 1453 (23%) with extra-pulmonary TB. The site of the initial phase of treatment was determined in 5790 patients: 1759 (30%) received DOT from guardians, 1465 (25%) from a health centre, 753 (13%) as out-patients from the hospital TB ward and 1813 (32%) remained in hospital. Eight-month treatment completion was 67% for smear-positive PTB patients, 51% for smear-negative PTB patients and 56% for extra-pulmonary TB patients. Two-month outcomes and 8-month treatment outcomes for all out-patient sites of supervision were satisfactory, except that a higher proportion of smear-positive PTB patients under guardian DOT failed to smear convert at 2 months. Over two-thirds of patients received ambulatory treatment out of hospital during the initial phase. CONCLUSION: The new treatment strategy, tested in five districts, was associated with a reduction in hospital bed days and satisfactory treatment outcomes. The results of these studies were vital in helping the National TB Control Programme make an informed decision about phased expansion of the strategy countrywide.