Abstract Purpose: Intracranial germ cell tumors (GCTs) are rare brain tumors affecting primarily adolescents and young adults. They usually arise at the pineal or suprasellar region, where surgical intervention is very challenging. There are two major subtypes, germinoma and nongerminomatous GCT (NGGCT). The treatment of intracranial GCT (iGCT) is based on histologic classification and therefore correct classification is vital to the improvement of the outcome. However, due to frequent mixed histology and relatively low sensitivity and specificity of tumor markers, it is a challenging task to accurately subclassify iGCT based on limited biopsy. Recently, aberrant expression of microRNA (miRNA) in various cancers has been reported and thus miRNA holds great promise as a biomarker for cancer diagnosis and classification as well as for understanding tumor biology. We did genome-wide miRNA expression profiling of 45 cases of iGCT and studied the utility of miRNA as diagnostic markers and classifier in iGCT. Methods: Tumor samples were collected from Texas Children's Cancer Center and three other collaborating institutions. Expression of 367 miRNA was measured by real time qPCR TaqMan assay. Differentially expressed miRNAs were used to train various supervised multivariate classification algorithms. Leave-one-out cross validation was used to estimate the overall accuracy, sensitivity and specificity of the classifier. We used a reiterative process to reduce the number of miRNAs in the classifier without sacrificing its discriminating power. Result: Using 33 fresh frozen and 23 formalin-fixed paraffin-embedded (FFPE) samples (11 cases with matched samples), each set of samples showed distinct clusters of germinomas and NGGCTs by unsupervised clustering. We confirmed that good quantification of miRNA expression was possible by using FFPE samples, which are generally more challenging to use for mRNA expression profiling. The classifier based on 7 miRNAs profiled in FFPE samples showed high sensitivity (87.5%) and specificity (100%) in distinguishing germinomas from NGGCTs. The same 7-miRNA classifier identified in FFPE series was able to distinguish the subtypes in fresh frozen samples with 93.3% sensitivity and 87.5% specificity. Similarly, the classifier based on 5 miRNAs profiled in fresh frozen samples was able to classify the subtypes in FFPE samples accurately. Many highly expressed miRNAs in iGCT are identical to those found in embryonal stem cells (ESCs). It is interesting to note that the miRNAs that are differentially expressed between two subtypes of iGCTs are similar to those that show significant changes in expression when undifferentiated ESCs undergo induced differentiation in vitro. Conclusion: We demonstrated that miRNA expression in FFPE specimens could be used in clinical diagnosis and classification of iGCT. Our results also suggest that iGCT and ESCs show similar miRNA profiles and could have implications on the cell of origin of iGCTs. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 102nd Annual Meeting of the American Association for Cancer Research; 2011 Apr 2-6; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2011;71(8 Suppl):Abstract nr 1144. doi:10.1158/1538-7445.AM2011-1144