Microarray analyses provide information on the relative expression levels of large numbers of gene products (transcripts). As such they have been widely used to examine differences in gene expression across a variety of samples such as tissues and life-cycle stages. Due to a previous lack of sequence data, microarray analyses have typically centred on the study of well-characterised model organisms. However, the recent availability of large sets of expressed sequence tags (ESTs) generated for the purpose of gene discovery offers the opportunity to consider designing and applying microarray technology to a larger and more diverse set of species. Here we outline the array-design process involving the generation of an optimised set of oligoprobes from a minimally redundant but maximally representative list of sequences from raw EST data. We illustrate these principles by showing how we designed and fabricated a high-density oligoarray for the rainbow trout, a non-model species for which large numbers of ESTs, and a non-redundant assembly is available. This approach brings array technology within the reach of all investigators, even those with limited budgets.