Genetic association, case-control studies are becoming a major instrument in the attempt to identify disease susceptibility markers of complex diseases. However, a major drawback of population-based studies of genetic association is the confounding effect of the population subdivision. We developed a statistic named T-value that estimates the differential transmission of marker alleles from heterozygous parents to the affected offspring, based on population data. Our method does not assume Hardy-Weinberg equilibrium and it can be used in very different population structures. A great advantage of this approach is that the genetic structure of the population can be assessed with a few unlinked loci and using classical population genetics theory (i.e. Wright's F-statistics). Four general models, assuming either one population with random mating, or one population without random mating, or several populations with random mating within them, or several populations without random mating within them, were developed to determine the behavior of the T-value under different mating conditions. Although a complete knowledge of the population structure is ideal to choose the best model, the simulations show that for a total inbreeding of 0.30 or less the last three models gave very similar estimates of the T-value. The model that assumed that total departure of Hardy-Weinberg proportions is due to population subdivision was the most robust under different scenarios of population structure. In sum, this study describes a novel procedure that can be used to identify the transmission of disease susceptibility markers in population-based studies.