DNA arrays are capable of profiling the expression patterns of many genes in a single experiment. After finding a gene of interest in a DNA array, however, labor-intensive gene-targeting experiments sometimes must be performed for the in vivo analysis of the gene function. With random gene trapping, on the other hand, it is relatively easy to disrupt and retrieve hundreds of genes/gene candidates in mouse embryonic stem (ES) cells, but one could overlook potentially important gene-disruption events if only the nucleotide sequences and not the expression patterns of the trapped DNA segments are analyzed. To combine the benefits of the above two experimental systems, we first created approximately 900 genetrapped mouse ES cell clones and then constructed arrays of cDNAs derived from the disrupted genes. By using these arrays, we identified a novel gene predominantly expressed in the mouse brain, and the corresponding ES cell clone was used to produced mice homozygous for the disrupted allele of the gene. Detailed analysis of the knockout mice revealed that the gene trap vector completely abolished gene expression downstream of its integration site. Therefore, identification of a gene or novel gene candidate with an interesting expression pattern by using this type of DNA array immediately allows the production of knockout mice from an ES cell clone with a disrupted allele of the sequence of interest.