abstract: The main aim of this paper is to describe a portfolio optimization problem under statedependent expected utility. Inspired by the recent results in Bjork, Murgoci and Zhou (2014), we analyze the mean–variance portfolio selection problem with statedependent risk aversion. We consider a continuous time Lévy model consisting of two assets, one stock price with dynamics of jumptype and a bank account with constant riskfree rate. Since this problem is time inconsistent we approach it within a game theoretic setting and look for subgame perfect Nash equilibrium strategies. Under this approach an extension of the standard dynamic programming equation to a system of nonlinear PDEs is needed. More precisely, using comparison theorems for BSDEs with jumps we arrive at tractable criteria for the solution of such games, in the form of a kind of nonMarkovian analogue of the HamiltonJacobiBellmanIsaacs (HJBI) equation.