Whole-genome association studies using single-nucleotide polymorphisms (SNPs) are the proposed method of choice for the identification of loci associated with complex diseases. In this report, we address the feasibility of generating high density SNP maps (with <100-kb spacing). As a pilot study, we concentrated on a 4-Mb region around the human APOE locus on chromosome 19, We compared the efficiency of SNP detection using YAC-based versus BAC/PAC-based maps, sequencing individual DNAs versus a pooled DNA sample, and we evaluated three different software applications for polymorphism detection. A total of 121 SNPs (25 in coding regions) were identified, The frequency of SNP detection was 1 SNP/1.1 kb of genomic sequence. From APOE to CALM3 (similar to 2 Mb), the average marker spacing was approximately 30 kb, Fifty-one SNPs were genotyped in five populations, and 10 SNPs showed an allele frequency differential greater than 0.5 between populations, Our results demonstrated that high-density SNP maps can be efficiently generated using existing technologies and that a genome-wide map with 60,000-100,000 SNPs is achievable in a reasonable time frame. (C) 1998 Academic Press.