With many different tissues and cell types, and possibly hundreds of different molecular interactions, the complexity of the immune system presents a formidable challenge to our understanding of immune disorders, most of which are caused by a combination of many genetic loci and environmental factors. Various ways of tackling this problem are discussed by articles in this issue. On page 483, Lawrence Steinman and Scott Zamvil argue that large-scale transcriptional analysis of immune-disease states, such as multiple sclerosis, is a powerful strategy for identifying new regulatory mechanisms and potential targets for treatment. Indeed, immunologists have long favoured the 'reverse' type of genetic analysis, which involves identifying a candidate gene by expression studies, such as those described by Steinman, then knocking it out to see if it causes an immune phenotype. The more traditional 'forward-genetic' approach is to start with a phenotype and then identify the gene responsible. As discussed by Vijay Kuchroo and colleagues on page 454, this longer-term approach has paid off with the identification of TIM1, an important asthma-susceptibility gene, and a highlight article on page 440 — Guilty by association — reports how the same type of analysis has led to the recent identification of susceptibility genes for type-1 diabetes. However, the availability of phenotypes is a real drawback to the forward-genetic approach — there are far fewer immune phenotypes than there are genes expressed by cells of the immune system — which is why some researchers are turning to chemical mutagenesis to create new phenotypes. On page 463, Mark Appleby and Fred Ramsdell discuss the power of this approach for dissecting the development and function of the immune system.