In this work we introduce two approximate duality approaches for vector optimization problems. The first one by means of approximate solutions of a scalar Lagrangian, and the second one by considering (C, ε)-proper efficient solutions of a recently introduced setvalued vector Lagrangian. In both approaches we obtain weak and strong duality results for (C, ε)-proper efficient solutions of the primal problem, under generalized convexity assumptions. Due to the suitable limit behaviour of the (C, ε)-proper efficient solutions when the error ε tends to zero, the obtained duality results extend and improve several others in the literature.Mathematics Subject Classification 90C48 · 90C25 · 90C46 · 49N15 L. Huerga: Researcher of Spanish FPI Fellowship Programme (BES-2010-033742).