This work has the ambitious aim of building a new Environmental Kuznets Curve (EKC), which accounts for non-homotheticity in green preferences and sharp nonlinearities in the impact of economic growth on the environment. Theoretically, our research is motivated by the fact that, if environmental quality is a good with low priority in the hierarchical scale, both the first- and the second-moments of the within-country distribution of income matters in shaping pollution impacts of growth. Particularly, for a given level of income per capita, a richer median voter will be more willing to approve more stringent environmental policies and thus a lower inequality is beneficial for the environment. With non-homothetic preferences, the beneficial environmental effect of reducing inequality emerges only for countries that are sufficiently rich. We test this hypothesis augmenting standard reduced form equation for EKC including the interaction between income per capita and the Gini coefficient. Our results for CO2, SO2 and NOx emissions corroborate our main hypothesis: reducing inequality is beneficial for the environment only for rich countries. Finally, we show that results are stronger for democracies.