The traditional, linear representation of the genetic information in populations cause a loss of data, as it can not fully represent sites where the sequences differ. Conversely, a graph may incorporate the variation, thereby providing a mean of utilizing more information in the analysis of genetic differences between populations. This thesis deals with the quantification of the differences between two population graphs. Three ways of measuring the distance between population graphs are presented. The first way counts unique variants and compares them to the total number of variants. The second calculates the graph edit distance. The third specifies two probability models regarding the genotype distribution at each variant, and then calculates the Bayes factor at each location. The three distance measures are tested on data describing variations in the human genome. Six populations of humans from distinct geographical areas are represented in the data. The distance measures seem to give similar conclusions about the relative distances among pairs of populations. In order to place the distances in a context, they are evaluated using permutation tests. The permutation tests report significant results for all pairs of populations except one. Additionally, this is the pair that is given the shortest distance by all distance measures. In general, the methods and ideas presented in this thesis allow more genetic information to be included in the study of relationships between groups. As such, it may prosper to become more accurate than traditional designs.