Theoretical models of the evolution of discrete phenotypes show that the most evolvable populations are composed of genotypes with intermediate levels of phenotypic robustness. This has been attributed to a special kind of epistasis, the analog of which for complex quantitative traits might not readily apparent. Here, with simulation models, I show that a variety of plausible kinds of quantitative genetic epistasis will do; as long as it increases cryptic genetic diversity and expected allele effect sizes are not too large. In fact, epistasis is not necessary, since cryptic genetic diversity can also accumulate via phenotypic plasticity. But with phenotypic plasticity, the mapping of phenotypic robustness to evolvability is sensitive to the nature and predictability of environmental variation. So, just as for discrete-traits, the robustness of quantitative traits can have complex effects on evolvability, and this depends on exactly how genetic diversity is hidden and revealed.