Although in Peru, Control Program of Tuberculosis has been successful in the reduction of incidence, morbidity and mortality, fatal cases are still observed. A longitudinal retrospective study was undertaken in a population on the outskirts of Lima between January 2000 and December 2005. Kaplan-Meier and Cox Regression analysis were used in this study. A total of 425 patients commencing tuberculosis treatment were included, with a case fatality ratio of 4.5%. Factors associated with mortality were: body mass index <18 kg/ m(2) at treatment beginning, HR = 4.89 (95%CI: 1.49 - 16.03) and HIV infection, HR=5.78 (95% CI: 1.11-29.99); whilst education level, HR = 0.28 (95% CI: 0.10 - 0.83) was associated with survival. Amongst this population commencing treatment for tuberculosis, body mass index and HIV infection were associated with an increased risk of death; in contrast, the education level was associated with improving in the survival.