<jats:p>Background: Surgical lung biopsy (SLB) is performed in patients with acute respiratory distress syndrome (ARDS); however, its clinical utility remains unclear. Objectives: We categorized the pathological diagnoses and investigated the predictive value for short-term mortality. Method: Three electronic databases (MEDLINE, the Cochrane Central Register of Controlled Trials, and ClinicalTrials.gov) were searched for the included studies. The QUADAS-2 was used to evaluate the risk of bias and its applicability. The types and populations of pathological diagnoses were investigated. The pooled sensitivity, positive likelihood ratio (LR+), negative likelihood ratio (LR−), and diagnostic odds ratio (DOR) were estimated at a fixed specificity. Hierarchical summary receiver operating characteristic curves were drawn. Results: A total of 16 studies that enrolled 758 patients were included. The pathological diagnoses were as follows: diffuse alveolar damage (DAD) 29.9%; infection 24.7%; interstitial lung disease 17.2%; malignancy 3.6%; cardiovascular disease 3.6%; drug toxicity 2.3%; connective tissue disease 2.2%; allergic disease 1.1%; and nonspecific diagnosis 15.4%. To predict short-term mortality, 13 studies that enrolled 613 patients used DAD as an index test and recorded a mortality rate of 56.9% (349 of 613 patients). A total of 3 studies that used index tests other than DAD were excluded. The pooled sensitivity, fixed specificity, LR+, LR−, and DOR were 0.46 (95% confidence interval [CI]: 0.29–0.56), 0.69, 1.48 (95% CI: 0.92–1.81), 0.78 (95% CI: 0.63–1.03), and 1.90 (95% CI: 0.89–2.86), respectively. Conclusions: SLB is unlikely to provide a specific diagnosis and should not be recommended for confirming DAD or predicting ARDS prognosis. </jats:p>