Genome-wide interaction studies is an extremely challenging problem in statistics, in which conventional methods are often inadequate in terms of both power and computational efficiency. An exhaustive search for genome-wide gene–gene interactions becomes feasible with modern cluster computing run on graphics processing units. However, the large number of tests accompanying the search raises a serious multiple testing problem. A way to overcome these limits is to apply a filtering step prior to the combinatorial method and to analyze only interesting single nucleotide polymorphisms selected based on a priori (defined by statistical evidence, genetic impact, or biological plausibility). The advantage of the filter approach is speed, and the disadvantage is that attributes with poor quality scores are disregarded. Genome-wide gene–environment interaction is less problematic computational demand compared with pairwise genome-wide gene–gene interaction. Accounting for gene–gene and gene–environment interactions is important for future strategies of diagnosis, prognosis, and management of human diseases and will bring new data regarding pathogenetic mechanisms for human complex diseases.