In our study, we aimed to profile a panel microRNAs (miRNAs) as potential biomarkers for the early diagnosis of pulmonary tuberculosis (PTB) and to illuminate the molecular mechanisms in the development of PTB. Firstly, gene expression profile of E-GEOD-49951 was downloaded from ArrayExpress database, and quantile-adjusted conditional maximum likelihood method was utilized to identify statistical difference between miRNAs of Mycobacterium tuberculosis (MTB)-infected individuals and healthy subjects. Furthermore, in order to assess the performance of our methodology, random forest (RF) classification model was utilized to identify the top 10 miRNAs with better Area Under The Curve (AUC) using 10-fold cross-validation method. Additionally, Monte Carlo Cross-Validation was repeated 50 times to explore the best miRNAs. In order to learn more about the differentially-expressed miRNAs, the target genes of differentially-expressed miRNAs were retrieved from TargetScan database and Ingenuity Pathways Analysis (IPA) was used to screen out biological pathways where target genes were involved. After normalization, a total of 478 miRNAs with higher than 0.25-fold quantile average across all samples were required. Based on the differential expression analysis, 38 differentially expressed miRNAs were identified when the significance was set as false discovery rate (FDR) < 0.01. Among the top 10 differentially expressed miRNAs, miRNA-155 obtained a highest AUC value 0.976, showing a good performance between PTB and control groups. Similarly, miRNA-449a, miRNA-212 and miRNA-132 revealed also a good performance with AUC values 0.947, 0.931 and 0.930, respectively. Moreover, miRNA-155, miRNA-449a, miRNA-29b-1* and miRNA-132 appeared in 50, 49, 49 and 48 bootstraps. Thus, miRNA-155 and miRNA-132 might be important in the progression of PTB and thereby, might present potential signatures for diagnosis of PTB.