Latent class (LCA) and cluster analysis (CLA) were utilized to identify trait loci for the Genetic Analysis Workshop 12 simulated disease. These techniques create non-overlapping subsets of concordant and discordant affected relative pairs based upon identity-by-descent (IBD) allele sharing at sequences of markers. Subgroups with a large proportion of affected pairs are used to identify markers in proximity to disease susceptibility loci. Both methods are model-free and make use of information from affected and unaffected subjects. In analyses performed without knowledge of the true disease model, LCA and CLA identified regions containing five of the seven trait loci.