DNA microarrays can be classified into oligonucleotides (Affymetrix) or long cDNAs (IncyteGenomics) based on the arrayed probes. Unfortunately, data are lacking on the comparison of these two popular global screening array systems. The present study was designed to assess the reliability of datasets generated by the two platforms from the same samples. We have already established a model for upregulation of a cluster of antioxidant responsive element (ARE)-driven genes in a human neuroblastoma cell line by treatment with tert-butylhydroquinone (tBHQ) for 8 and 24 h. HuGene FL (Affymetrix), U95 Av2 (Affymetrix), and UniGem V 2.0 (IncyteGenomics) were chosen to do the comparative study on 8- and 24-h samples. The Affymetrix data generated from U95Av2 chips demonstrated that the mRNA of 218 (2.3% of total clones) genes was increased after 8 h of tBHQ treatment. This list included most of the known ARE-driven genes, and nine selected genes showed a high consistency with RT-PCR results. IncyteGenomics called four genes increased and no genes were decreased. These same four genes were also called by the Affymetrix microarray. The sensitivity (fluorescence intensity) and specificity (fold) were very different for selected genes when comparing the two platforms. Cross-hybridization was shown to partially contribute to the discrepancies of the data generated by the two platforms. According to our results, the data generated from oligonucleotide microarrays is more reliable for interrogating changes in gene expression than data from long cDNA microarrays.