An important goal of human genetics and genomics is to understand the complete spectrum of genetic variation across a specific human haplotype. By combining information from a dense SNP map with fosmid end-sequence pairs (ESPs) aligned to the human genome reference sequence, we have developed a simple method to resolve human haplotypes using a previously developed clone resource. By partitioning ESPs into either haplotype, we have generated a haplotype-specific clone map for eight diploid genomes (four Yoruba African and four non-African samples). On average, 59% of each haploid genome is covered by haplotype-assigned clones with an N50 length of 110 kbp. By comparing this clone-based haplotype map against HapMap phased data sets, we estimate an error rate of 0.71% when trio information is available and 6.6% in its absence. We present these data in the form of an interactive browser that allows clones corresponding to specific haplotypes to be recovered and sequenced within these eight human genomes. As an example, we sequenced 165 fosmid clone inserts to generate 6.8 Mbp of sequenced haplotypes, and demonstrate its utility in uncovering phase-switching errors and for the discovery of novel SNPs especially in Asian and African samples. We discuss the potential application of this resource in understanding the pattern of genetic variation in complex regions of the genome that may not be adequately resolved by next-generation sequencing technology or SNP haplotype imputation.[Supplemental material is available online at www.genome.org.]Next-generation high-throughput sequencing technologies promise to greatly accelerate the pace of genomics research (Mardis 2008). Coupled with methods of "genomic selection," these technologies allow the acquisition of sequence from a targeted subset of the genome Okou et al. 2007;Porreca et al. 2007). The information obtained from these methods, however, is a mixture of the two haplotypes represented in a given sample. Although statistical approaches and pedigree information allow for the imputation of phased haplotypes (Stephens et al. 2001;Stephens and Donnelly 2003), the acquisition of long, contiguous stretches of sequence derived from a single haplotype remains an important goal for human genetics research. Recently, an effort to systematically map and sequence structural variants using a fosmid-based clone-end approach was launched with the goal of resolving structurally variant haplotypes at the sequence level and ultimately incorporating alternative human haplotypes into a more comprehensive reference assembly (Eichler et al. 2007). These clone libraries were derived from diploid genomic DNA from individuals studied as part of the HapMap project (The International HapMap Consortium 2005). Here, by combining the mapping information associated with these clones with HapMap single nucleotide polymorphism (SNP) genotypes, we define a haplotype-specific physical clone map for these eight individuals (corresponding to 16 haploid genomes). As a result, we directly assign a subset...