Referral public services are essential concerning TB-HIV co-infected patients, but they do not exempt from the geographical and social complicators where they are inserted in. Retrospective data analysis of the TB Surveillance and Information System (TBWEB) regarding co-infected TB/HIV inpatients discharged from July 2008 to June 2009. Were included 209 consecutive cases. From those, 47 (22.5%) were not notified to TBWEB. From the 162 (77.5%) notified cases under TB treatment initiated or sustained while in hospital, 60 (37.0%) presented pulmonary-TB, other 60 (37.0%) extrapulmonary-TB and 42 (26.0%) mixed forms. 99 cases (61.1%) were considered as a new TB case, 29 (17.9%) were re-treatments after default, 5 (3.1%) were re-treatments after failure and 29 (17.9%) had been relapsed. There was no culture record in 60 (37.0%) cases. In other 102, culture was positive in 68 (42.0%). Sensitivity test was performed in 54 of these 68 (79.4%), 37 stains were all-sensitive (68.5%), 5 (9.3%) were monoresistant, and 12 (22.2%) were multiresistant to TB-drugs. Cure was achieved in 45.7% of the 162 notified cases; default in 29.6%; failure in 3.7%; change of diagnosis in 3.7% (due to non-tuberculosis mycobacteria); 3.7% remain under treatment. Were notified 22 deaths (13.6%). The high rates of under-reporting, cultures and sensitivity tests not performed and, most of all, the low rate of cure urges practical and effective measures to improve this reality. A deeper analysis of these data are mandatory for better understanding and searching for solutions. A stronger commitment to DOTS Strategy rises as a possible answer.