Analysis of gene expression data sets is a potent tool for gene function prediction, cis-element discovery, and hypothesis generation for the model plant Arabidopsis thaliana, and more recently for other agriculturally relevant species. In the case of Arabidopsis thaliana, experiments conducted by individual researchers to document its transcriptome have led to large numbers of data sets being made publicly available for data mining by the so-called "electronic northerns," co-expression analysis and other methods. Given that approximately 50% of the genes in Arabidopsis have no function ascribed to them by "conventional" homology searches, and that only around 10% of the genes have had their function experimentally determined in the laboratory, these analyses can accelerate the identification of potential gene function at the click of a mouse. This chapter covers the use of bioinformatic data mining tools available at the Bio-Array Resource ( http://www.bar.utoronto.ca ) and elsewhere for hypothesis generation in the context of seed biology.