Epistasis, commonly defined as the interaction between multiple genes, is an important genetic component underlying phenotypic variation. Due to large combinatorial search spaces, most current methods face enormous computational challenges and suffer from low statistical power after multiple test correction. Here, we present an alternative strategy for mapping epistasis: instead of identifying explicit genetic pairs, we focus on mapping variants that have non-zero marginal epistatic effects —the combined pairwise interaction effect between a given variant and all other variants. By testing marginal epistatic effects, we identify candidate variants that are involved in epistasis without the need to identify the exact partners, thus alleviating much of the statistical and computational burden associated with standard epistatic mapping procedures. We refer to our method as the “MArginal ePIstasis Test”, or MAPIT. With simulations and real data examples, we show how MAPIT can be used to produce calibrated test statistics under the null, facilitate the detection of pairwise epistatic interactions, and estimate the broad-sense heritability of complex traits.