Background: Determine TB-LAM Ag (LAM) is a point of care test developed to diagnose tuberculosis (TB). The aim of this study was to evaluate the diagnostic performance of LAM in people living with HIV using Brazilian public health network algorithm for TB diagnosis. Methods and findings A cross-sectional study design was used to enroll 199 adult patients in two sites in Rio de Janeiro and two in São Paulo. The study enrolled HIV-infected patients with CD4 counts ≤ 200 cells/mm3 (in the Alere PIMA CD4 assay at study screening), patients coughing for at least 2 weeks or presenting a chest radiography suggestive of TB. LAM, in conjunction with sputum smear microscopy or Xpert MTB/RIF (Xpert) as compared to Mycobacterium tuberculosis culture, which was used as a reference standard. TB prevalence was 24.6%. Overall accuracy of LAM was 79.9% (73.8%-84.9%), positive and negative predictive values were 62.2% (46.1%-75.9%) and 84% (77.5%-88.8%), respectively. The overall LAM sensitivity was 46.9% (33.7%-60.6%) and specificity was 90.7% (84.9%-94.4%). The best performance of LAM was observed among patients with CD4 counts ≤ 50 cells/mm3 (sensitivity = 70.4% and specificity = 85.9%). When 2 respiratory smears were used in conjunction with LAM, sensitivity increased 22%, as compared to just 2 smears. Furthermore, LAM when used in conjunction with two respiratory smears, was as sensitive as compared to a single one. However, no improvement in TB diagnosis occurred when LAM was used with Xpert as compared to Xpert alone. Among 14 LAM false positive tests, Non-Tuberculosis Mycobacteria were isolated in three cases. Conclusion: LAM is a point of care test that increased TB diagnosis in immunosuppressed HIV-infected patients when used in conjunction with smear microscopy, but not when used with Xpert in Brazilian public health network sites. Use of LAM test should be considered in settings where immunosuppressed HIV patients need rapid TB diagnosis.