A major inconvenience in cloning a dominant mutation in Saccharomyces cerevisiae is the requirement for the construction of a genomic library from the mutant strain (1). To alleviate the need for library construction, we present an alternative that is based on the ability of yeast cells to carry out precise and efficient homologous recombination (2). In principle, a dominant mutation may be cloned by co-transforming (3) a genomic library of a wild-type strain with genomic DNA derived from the mutant strain and selecting for transformants with the mutant phenotype. These should represent the following events: (i) Recombination between the genomic DNA fragment from the mutant strain, carrying the mutant gene, and the respective locus in the genome of the wild-type recipient strain. (ii) Recombination between the same DNA and a library plasmid that contains the gene of interest, if they coexisted within the same cell. (iii) A spontaneous mutation resulting in the mutant phenotype. A fourth possibility where a few additional copies of a wild-type gene would result in the same mutant phenotype, is considered unlikely and would be a caveat even for the classical approach. Co-segregation of the mutant phenotype with the plasmid should facilitate the selection of the second class events that eventually result in the isolation of the mutant gene of interest. We initially exploited the possibility of correcting a defect in a known gene carried on a centromeric plasmid, by applying the above outlined scheme of recombination between a plasmid and a genomic DNA fragment. A gcn2 , ura3-52 strain (4), sensitive to l0 mM of the drug 3-aminotriazole (3-AT), was co-transformed with 20 g of a URA3 marked plasmid (YCp50) harbouring a defective GCN2 gene, along with 50 g of genomic DNA (5) from the wild-type strain S288C. The GCN2 gene on the plasmid had been inactivated by digesting, filling in and religating the unique XhoI site within the GCN2 ORF (Fig. 1). Out of 20 000 primary Ura+ transformants, seven were resistant to 3-AT after replica-plating on minimal plates containing l0 mM 3-AT. In six cases this resistance was shown to be plasmid borne following eviction of the plasmid by treating the transformed cells with 1.0 mg/ml 5-Fluoroorotic acid (5-FOA; ref. 2). The plasmids from these six transformants were recovered (6) and checked for the fidelity of the correction, i.e. restoration of the XhoI restriction site. The results are shown in Figure 1. In this experiment, a correction could not have resulted from exchange of information between the plasmid, Figure 1. Repair of a gcn2 allele following co-transformation with genomic DNA fragments. (Top) Diagrammatic representation of the GCN2 gene locus showing the positions of the unique XhoI and the flanking BamHI sites. (Bottom) Restriction analysis of GCN2 bearing plasmids. Lane 1, HindIII digest of phage DNA, used as size standard. Lane 2, the YCp50 plasmid carrying the gnc2 mutant gene generated by destroying the XhoI site, digested with BamHI and XhoI. Lane 3, BamHI/XhoI double digest of a plasmid recovered from a 3-AT sensitive transformant. No restoration of the XhoI is evident. Lanes 4–9, BamHI/XhoI double digests of the plasmids recovered from the 3-AT resistant transformants. In all cases the XhoI site has been restored. Lane 10, BamHI/XhoI control digest of YCp50 carrying the wild-type GCN2 gene.