Microarray technology is a current approach for detecting alterations in the expression of thousands of genes simultaneously between two different biological conditions. Genes of interest are selected on the basis of an obtained p-value, and, thus, the list of candidates may vary depending on the data processing steps taken and statistical tests applied. Using standard approaches to the statistical analysis of microarray data from individuals with Autism Spectrum Disorder (ASD), several genes have been proposed as candidates. However, the lists of genes detected as differentially regulated in published mRNA expression analyses of Autism often do not overlap, owed at least in part to (i) the multifactorial nature of ASD, (ii) the high inter-individual variability of the gene expression in ASD cases, and (iii) differences in the statistical analysis approaches applied. Game theory recently has been proposed as a new method to detect the relevance of gene expression in different conditions. In this work, we test the ability of Game theory, specifically the Shapley value, to detect candidate ASD genes using a microarray experiment in which only a few genes can be detected as dysregulated using conventional statistical approaches. Our results showed that coalitional games significantly increased the power to identify candidates. A further functional analysis demonstrated that groups of these genes were associated with biological functions and disorders previously shown to be related to ASD.