Objective The study aimed to analyze the main clinical risk factors of death in patients with pulmonary tuberculosis secondary respiratory failure, and build a quantitative risk prediction model to provide reference for clinical practice. Method A total of 105 patients with pulmonary tuberculosis secondary respiratory failure to Danyang Hospital affiliated to Kangda College of Nanjing Medical University from March 2019 to June 2022 were retrospectively summarized. According to the clinical outcome, they were divided into death group (n=30, 28.6%), and survival group (n=75, 71.4%). The general clinical datas of the two groups were analyzed, including gender, age, sputum smear positive rate, sputum culture positive rate, nucleic acid quantitative detection Xpert MTB/RIF positive rate, comorbid diseases, complications, tuberculosis classification, admission pulmonary function (forced vital capacity, tidal volume, residual gas), admission arterial blood gas analysis[arterial partial pressure of oxygen (PaO2), partial pressure of carbon dioxide (PaCO2), oxygen saturation (SaO2), PaO2/FiO2], APACHEⅡ score, C-reactive protein (CRP), white blood cell count, neutrophil/lymphocyte ratio(NLR), mean arterial pressure (MAP), albumin when arriving at hospital; Then, multivariate Logistic regression analysis was used to screen the main risk factors those may affect the death outcome, and the risk prediction model was constructed. Finally, the receiver operating curve (ROC curve) was used to evaluate the diagnostic efficacy. Result Compared with the survival group, the complications in the death group were increased, the hospital forced vital capacity was decreased, PaO2/FiO2 was decreased, NLR was increased, and albumin was decreased (P<0.05). In the death group,the bacterial resistance rate and invasive ventilation rate were increased, ventilation time was prolonged, treatment 3d-PaO2/FiO2 was decreased, APACHEⅡ score was increased, CRP and NLR were increased, albumin was decreased (P<0.05). Logistic regression model showed that PaO2/FiO2, NLR, albumin, APACHEⅡ score and CRP after three-day treatment were the main risk factors to death (P<0.05). The risk prediction model Y=2.563-0.248×(PaO2/FiO2)+0.166×NLR-0.318×albumin-0.074×(APACHEⅡ score)+0.396×CRP. ROC analysis showed that the accuracy of Y value for predicting death was 0.856, sensitivity was 82.6%, specificity was 75.5%, and cut-off value was 0.869. Conclusion Pulmonary tuberculosis secondary respiratory failure has a high in-hospital fatality rate. Lower of PaO2/FiO2 and albumin, higher of NLR, APACHEⅡ score and CRP after three-day treatment may be the main risk factors affecting death outcome, which has good accuracy and application value to build a quantitative risk prediction model for evaluating death.