A high-throughput two-step PCR strategy for the identification of selected genes from a BAC library derived from hexaploid wheat (16,974 Mbp) is described. The screen is based on the pooling of DNA from BAC clones into 675 "superpools" arrayed in a three-dimensional configuration. Each BAC clone is represented in three superpools to allow the identification of candidate 384-well plates of clones after the first round of PCR; identification is facilitated by an associated Perl script. A second round of PCR detects the specific BAC clone within the candidate plate that corresponds to the gene of interest. Thus, a single copy of the target gene can be identified from the library of over 700,000 clones (approximately 5 genome equivalents) by assaying only three 384-well plates. The pooling strategy was validated by screening the library with primers specific for the reduced height (Rht-1a) gene. Using relatively stringent selection criteria, 13 Rht-containing clones were identified from 17 candidate plates, and sequence analysis of the amplified products showed that all three Rht homoeologues were represented. Furthermore, the method confirmed the estimated coverage of the BAC library. Thus, this methodology allows the rapid and cost-effective identification of genes, and their homoeologues, from large-insert libraries of complex genomes such as hexaploid wheat.