The large number of single nucleotide polymorphisms (SNPs) available provides a powerful molecular resource for identifying complex genetic interactions associated with complex traits or diseases but also presents unprecedented data analysis challenges. In this work we developed new quantitative genetics methods and parallel computing tools to detect complex interactive SNP effects underlying complex traits or diseases using genome-wide association studies (GWAS). The new quantitative genetics methods allow detection of novel interactions between genes, sex and environment including second order and third order gene-gene, gene-sex, gene-environment interactions, where each gene may have additive, dominance or parent-of-origin effects. The parallel computing tools allow such complex analysis to be conducted in a timely manner for any large scale GWAS and can be scalable to meet growing data analysis challenges in the future. The analytical and computing methods were applied to the analysis of a Holstein cattle GWAS data set and the Framingham Heart Study (FHS) data. Significant epistasis and single-locus effects were detected affecting human cholestoral levels and dairy production, fertility and body traits. The analytical methods and computing tools will significantly facilitate the discovery of complex mechanisms underlying phenotypes using GWAS.