This study investigated the relationship between gene expression and disease based on the expression profiles of tissue-specific genes, with the aim of discovering the candidate genes associated with disease risk as diagnostic markers. The gene-expression profiles of approximately 20,000 genes from 4 anaplastic astrocytomas (AAs), 7 glioblastoma multiformes (GBMs), and 1 nontumor brain (NB) were analyzed by in situ-synthesized 60-mer oligonucleotide microarray. The signal intensity of each feature was measured by laser scanner, and gene expression was quantified as the tumor/NB intensity ratio. Gene expression was defined as having increased or decreased when the ratio was >or=1.5 or <or= 0.67, respectively. In comparison with NB, 20 genes showed increased expression, and 32 showed decreased expression in AA, while 20 genes showed increased expression, and 164 showed decreased expression in GBM. This research indicates that changes in expression levels of 4 genes in AA and 6 genes in GBM may be useful diagnostic markers. Quantitative reverse transcription-polymerase chain reaction also supported the microarray results for the 6 selected genes. In conclusion, microarray images of excellent quality are of high data reproducibility using entirely standard procedures, and these genes may be associated with the molecular events leading to AA or GBM.