Although there are several methods for genotyping previously identified single nucleotide polymorphisms (SNPs), there is a paucity of approaches for high-throughput scanning for unknown variations. Mismatch repair detection (MRD) utilizes a bacterial mismatch repair system in vivo to detect sequence variants in human DNA samples. We describe modifications in MRD that allow a high degree of parallel processing, and use this modified version to accurately scan for variations in 35 different human DNA fragments simultaneously. MRD's potential for high-throughput scanning can be used to identify new SNPs and to comprehensively compare sequences between patients and controls for identifying disease susceptibility alleles.