The decomposition of genetic variance into additive, dominance, and epistatic components is a common procedure in quantitative genetics. Yet, the interpretation of this variance partition is not trivial, especially concerning nonadditive components. In this chapter, we compile various uses of variance partitioning from published analyses, new simulations, and theoretical examples. We show ways in which advanced genetic modeling facilitates the analysis of data through variance partitioning, focusing on the natural and orthogonal interactions (NOIA) model. We also discuss how epistasis and epistatic variance may influence the outcome of selection, a topic that is still a matter of debate among quantitative and evolutionary geneticists.