Identifying gene and environment interaction (G × E) can provide insights into biological networks of complex diseases, identify novel genes that act synergistically with environmental factors, and inform risk prediction. However, despite the fact that hundreds of novel disease-associated loci have been identified from genome-wide association studies (GWAS), few G × Es have been discovered. One reason is that most studies are underpowered for detecting these interactions. Several new methods have been proposed to improve power for G × E analysis, but performance varies with scenario. In this article, we present a module-based approach to integrating various methods that exploits each method's most appealing aspects. There are three modules in our approach: (1) a screening module for prioritizing Single Nucleotide Polymorphisms (SNPs); (2) a multiple comparison module for testing G × E; and (3) a G × E testing module. We combine all three of these modules and develop two novel “cocktail” methods. We demonstrate that the proposed cocktail methods maintain the type I error, and that the power tracks well with the best existing methods, despite that the best methods may be different under various scenarios and interaction models. For GWAS, where the true interaction models are unknown, methods like our “cocktail” methods that are powerful under a wide range of situations are particularly appealing. Broadly speaking, the modular approach is conceptually straightforward and computationally simple. It builds on common test statistics and is easily implemented without additional computational efforts. It also allows for an easy incorporation of new methods as they are developed. Our work provides a comprehensive and powerful tool for devising effective strategies for genome-wide detection of gene-environment interactions.