The choice of an optimal marker strategy while analysing intragenic SNPs is presently of crucial importance, given the increasing amount of available data. Classical case/control association studies or family based association tests such as the TDT are very popular. However, as these methods are not able to analyse multiple markers simultaneously, different extensions have been proposed in order to use multiple markers. In the present study, the efficiency of five family based haplotypic methods to detect the role of candidate genes is evaluated and compared between them and with the classical single point TDT. Simulations of intragenic SNP maps are performed in recently founded populations. One or several SNPs are assumed to be the functional polymorphisms following different genetic models. Different modes of SNP combinations underlying the genetic susceptibility (epistasis or heterogeneity) are considered. Whereas haplotypic methods perform better in situations of heterogeneity, the TDT remains the most powerful approach in epistasis models as long as the marginal effect of one the SNPs involved in the susceptibility remains important. Haplotypic methods perform better than the TDT when the marginal effect of each SNP is small. Given the similar characteristics of intragenic LD in both old large populations and recently founded populations, in particular the weak correlation between LD and distance, our results are not likely to be specific to founder populations and can be generalized.