Meta-analysis of microarray data from four published cold stress experiments in Arabidopsis thaliana were performed to detect differentially expressed genes (DEGs).The results were compared with those obtained by the separate analyses of the individual experiments. The results showed that most of those “more reliable” genes, which were detected in two or more experiments, were detectable in the meta-analysis, whereas merely a quarter of those “less reliable” genes, which were detected only in a single experiment, were detectable in the meta-analysis. This suggested that meta-analysis was suitable for detecting “more reliable” DEGs. It was noticeable that 9% of the genes detected by the meta-analysis, among which 6 genes were directly related to the response to cold stress, were not detected by the separate analyses, indicating that meta-analysis could detect some DEGs neglected by separate analyses. Gene ontology (GO) and promoter motif analyses suggested that the DEGs detected by the meta-analysis were indeed closely related to cold response.