Background: Increased morbidity and mortality of tuberculosis have been blamed on neglect of the human dimension of tuberculosis control. One of such factors included in human dimension is non-compliance, a behavioural parameter, which has led to the emergence of multi-drug resistant tuberculosis, and poor treatment outcome. Objective: To explore the impact of directly observed therapy (DOT) on compliance and the factors influencing it. Design: A retrospective study. Setting: Chest clinic of Obafemi Awolowo University Teaching Hospitals Complex, Ile-Ife, Nigeria. Methods: Directly observed therapy (DOT) was employed. Records of the socio-demographic characteristics, treatment categories, complications developed, results of investigations, level of compliance and treatment outcome for the patients were kept. The data for the patients seen between May 1996 and April 1997 were retrieved and analysed. Those that complied were compared with those that did not comply. Results: One hundred and ninety nine patients comprising ninety one males and 108 females, were seen during the period .They were mostly between the ages of 16 years and 45 years(mean ± SD = 31.7±14.98). One hundred and fifty eight (73%) complied and all of them were cured. The only factor that significantly influenced rate of compliance was proximity to the chest clinic. Conclusion: DOT improves the rate of compliance. No socio-demographic factors considered significantly influenced the rate of compliance under DOT, and as such they are not reliable predictive factors. Locating chest units in the existing primary health care facilities will improve the rate of compliance with antituberculosis therapy. More attention should be paid to behavioural aspect of tuberculosis control.