Laryngeal cancer (LC) is a common poor prognostic head and neck cancer. MiRNAs are promising biomarkers for diagnosis of LC. This study aims to evaluate miRNA-190, miRNA-588 and miRNA-802 expression levels and their significance as molecular biomarkers for LC. Furthermore, a prediction model for LC late stage was created to discover patients at high risk of late stage development. Quantitative real-time polymerase chain reaction (RTqPCR) was used to investigate the expression levels of these genes in sixty cancerous tissue specimens and adjacent thirty healthy specimens. The demographic and clinical data, along with the TNM staging of LC patients were also investigated. Significant down-regulation of MiRNA-588 and miRNA-802 in LC (P<0.001) showed a tumor suppressor role with the possibility of their usage as biomarkers differentiating between cancerous and healthy tissues. ROC curve analysis confirmed the diagnostic ability of miRNA-588 and miRNA-802 to discriminate between cancerous and healthy tissues. In the late stage, decreased expression level of miRNA-802 was only detected when compared with early stage (P=0.004) indicating that miRNA-802 might play a protective role against progression to late stages. The multivariate analysis revealed that only stridor (P<0.001) and Cervical lymphadenopathy (P=0.009) might serve as independent predictors of LC late stage.