The precise characterization of linkage disequilibrium (LD) patterns through the genome is important for different purposes that range from the better understanding of the demographic history of populations to the identification of disease associated genes. Even though most of these applications are effective when LD is calculated at pairs of biallelic loci, specific regions of the genome like the HLA region display a high genetic diversity with multiple alleles per loci. Unfortunately, LD measures designed for the first situation cannot be easily extended to the second, and many different new measures have been proposed for pairs of multiallelic loci without a clear idea of which could be better than the others. In addition, most measures of LD require of knowledge regarding the haplotypic phase, whose estimation becomes challenging at multiallelic systems. During this project, these problems are extended with further explanations and confronted from different perspectives. First, multiple measures of interallelic LD at pairs of multiallelic loci are compared between them to find potential advantages and disadvantages. The extension of two measures of overall LD to interallelic LD is also evaluated, with clear disadvantages compared to more commonly used measures. Then, the haplotypic phase estimation performance of the EM-algorithm is evaluated in comparison to the real haplotypic phase, obtained from family data. As multiple populations and sample sizes are employed, it has been possible to find association between some population parameters and the accuracy of phase estimation by the EM-algorithm. Finally, a new methodology to apply on usual measures of interallelic LD is proposed in order to improve the description of the situation when other measures are used and identify haplotypes with the highest relative LD at the loci pair.