PCR product-based gene disruption has greatly accelerated molecular analysis of Saccharomyces cerevisiae. This approach involves amplification of a marker gene (e.g., URA3) including its flanking regulatory (promoter and polyadenylation) regions using primers that include at their 5' ends about 50 bases of homology to the targeted gene. Unfortunately, this approach has proved less useful in organisms with higher rates of non-homologous recombination; e.g., in the yeast Candida glabrata, desired recombinants represent < or =2% of transformants. We modified the PCR-based approach by eliminating marker-flanking regions and precisely targeting recombination such that marker expression depends on the regulatory sequences of the disrupted gene. Application of this promoter-dependent disruption of genes (PRODIGE) method to three C. glabrata genes (SLT2, LEM3, and PDR1) yielded desired recombinants at frequencies of 20, 31, and 11%, the latter representing a weakly expressed gene. For Candida albicans LEM3 and RHO1, specificity was 79-95% for one or both alleles, >sixfold higher than the published results with conventional PCR-based gene disruption. All 5 C. glabrata and C. albicans mutants had predicted phenotypes of calcofluor hypersensitivity (slt2Delta and RHO1/rho1Delta), cycloheximide hypersensitivity (pdr1Delta), or miltefosine resistance (lem3Delta and lem3Delta/lem3Delta). PRODIGE application to the S. cerevisiae PDR5 gene in strains with and without the Pdr1-Pdr3 transcriptional activators of this gene confirmed that transformant yield and growth rate depend on promoter strength. Using this PDR5 promoter-URA3 recombinant, we further demonstrate a simple extension of the method that yields regulatory mutants via 5-fluoroorotic acid selection. PRODIGE warrants testing in other yeast, molds, and beyond.