Tumor microenvironments present significant barriers to anti-tumor agents. Molecules involved in multicellular tumor microenvironments, however, are difficult to study ex vivo. Here, we generated a matrix-free tumor spheroid model using the NCI-H226 mesothelioma cell line and compared the gene expression profiles of spheroids and monolayers using microarray analysis. Microarray analysis revealed that 142 probe sets were differentially expressed between tumor spheroids and monolayers. Gene ontology analysis revealed that upregulated genes were primarily related to immune response, wound response, lymphocyte stimulation and response to cytokine stimulation, whereas downregulated genes were primarily associated with apoptosis. Among the 142 genes, 27 are located in the membrane and related to biologic processes of cellular movement, cell-to-cell signaling, cellular growth and proliferation and morphology. Western blot analysis validated elevation of MMP2, BAFF/BLyS/TNFSF13B, RANTES/CCL5 and TNFAIP6/TSG-6 protein expression in spheroids as compared to monolayers. Thus, we have reported the first large scale comparison of the transcriptional profiles using an ex vivo matrix-free spheroid model to identify genes specific to the three-dimensional biological structure of tumors. The method described here can be used for gene expression profiling of tumors other than mesothelioma.