Tuberculosis (TB) is an increasing global problem, despite effective drug therapies. Access to TB therapy during conflict situations has not been studied. To determine the effect of irregular TB treatment due to an armed conflict in Guinea-Bissau, West Africa. Ongoing retrospective cohort study conducted in the capital city of Bissau among 101 patients with TB who received irregular or no treatment during the civil war (war cohort; June 7-December 6, 1998) and 108 patients with TB who received treatment 12 months earlier (peace cohort; June 7-December 6, 1997) and comparison of an additional 42 patients who had completed treatment before June 6, 1998, and 69 patients who had completed treatment before June 6, 1997. Mortality rates, compared by irregular (war cohort) vs regular (peace cohort) access to treatment, by intensive vs continuation phase of treatment, and by those who had previously completed treatment for TB. Irregular treatment was associated with an increased mortality rate among patients with TB. The mortality rate ratio (MR) was 3.12 (95% confidence interval [CI], 1.20-8.12) in the war cohort, adjusting for age, sex, human immunodeficiency virus (HIV) infection, residence, and length of treatment. Each additional week of treatment before the war started increased probability of survival by 5% (95% CI, 0%-10%). In the intensive phase of treatment, the adjusted MR was 3.30 (95% CI, 1.04-10.50) and in the continuation phase it was 2.26 (95% CI, 0.33-15.34). Increased mortality among the war cohort was most marked in HIV-positive patients, who had an adjusted MR of 8.19 (95% CI, 1.62-41.25). Mortality was not increased in HIV-positive or HIV-negative patients who had completed TB treatment when the war started. Interruption of treatment had a profound impact on mortality among patients with TB during the war in Guinea-Bissau. Regular treatment for TB was associated with significantly improved survival for HIV-infected individuals. In emergencies, it is crucial to ensure availability of TB drugs.