We describe a haplotype clustering approach for localising a disease mutation within a fixed genomic region, which supplements tagging SNP (tSNP) information with (external) information on linkage disequilibrium. By applying our method to simulated data based on the coalescent, and on real haplotype data, we demonstrate that there are situations where significant gains can be made by incorporating tagged SNPs into the analysis. The issues we explore are important not only to these types of studies, but also to studies that select tSNPs based on (external) HapMap phase II data, and those that use genome-wide markers.