MicroRNA-134 (miR-134) has been demonstrated to be dysregulated in glioma tissues. However, its clinical significance in this tumor type has not been fully elucidated. This study was designed to explore the association of miR-134 expression with clinicopathological features and prognosis in human glioma. Using real-time quantitative polymerase chain reaction, miR-134 expression was detected in 162 glioma specimens with various World Health Organization (WHO) grades and compared to the expression in 36 normal brain tissue samples. Glioma tissues exhibited significantly reduced expression of miR-134 (mean 2.15±standard deviation 0.82 versus 4.37±1.16, p<0.001) compared with normal brain tissues. In addition, miR-134 expression was notably associated with WHO grade (p<0.001) and Karnofsky Performance Scale score (KPS; p=0.02) in glioma tissues. Low miR-134 expression occurred more frequently in glioma tissues with high WHO grades and low KPS scores. In univariate analysis, both progression-free survival (PFS) and overall survival (OS) of glioma patients with low miR-134 expression were significantly shorter than those with high miR-134 expression (both p<0.001). Additionally, glioma patients with high WHO grades, low KPS scores and subtotal resection attained significantly poorer PFS (p<0.001, 0.02 and 0.01, respectively) and OS (p<0.001, 0.01 and 0.01, respectively). In multivariate analysis, miR-134 expression, WHO grade, KPS score and extent of resection were identified as the independent prognostic factors for both PFS and OS. Collectively, our data prove that the reduced expression of miR-134 may predict aggressive progression and poor prognosis in human gliomas. miR-134 may represent both a prognostic marker and a novel therapeutic target for this malignancy.