We investigate the lattice QCD Dirac operator with staggered fermions at temperatures around the chiral phase transition. We present evidence of a metal-insulator transition in the low lying modes of the Dirac operator around the same temperature as the chiral phase transition. This strongly suggests the phenomenon of Anderson localization (localization by destructive quantum interference) drives the QCD vacuum to the chirally symmetric phase in a way similar to a metalinsulator transition in a disordered conductor. We also discuss how Anderson localization affects the usual phenomenological treatment of phase transitions a la Ginzburg-Landau.PACS numbers: 72.15. Rn, 71.30.+h, 05.45.Df, One of the most important features of the infrared limit of the strong interactions is the spontaneous breaking of the approximate chiral symmetry. The order parameter associated with this spontaneous chiral symmetry breaking (SχSB) is the chiral condensate, ψ ψ , which in the absence of SχSB would vanish as the quark mass goes to zero. In nature the lightest quarks are not massless so a nonzero condensate is expected even in a free theory. However the small quark mass can only account for a small percentage of the chiral condensate, the rest has its origin in the strong non-perturbative color interactions of QCD. Although lattice simulations have already provided overwhelming evidence that SχSB is a feature of QCD it is still highly desirable to understand its origin in more simple terms.Simplified models of QCD where gauge configurations are given by instantons have played a leading role in the description of the SχSB [1,2,3]. Instantons [4], originally introduced in QCD by t'Hooft [5] to solve the so called U (1) problem, are classical solutions of the Euclidean Yang-Mills equations of motion. Their relation with the SχSB stems from the fact that the QCD Dirac operator has an exact zero eigenvalue in the field of an instanton. In the QCD vacuum, these zero modes coming from different instantons mix together to form a band around zero. It turns out the spectral density, ρ(λ), of the Dirac operator in this region is directly related to the chiral condensate through the Banks-Casher relation [6],where V is the space-time volume. Based on this result it was shown that the instanton contribution was capable of producing a nonzero chiral condensate with a value close to the phenomenological one [2, 3] (see [7] for a detailed review). For sufficiently high energies the non-abelian gauge interaction of QCD is weak (asymptotic freedom) and chiral symmetry is restored. This poses an interesting question: How is the chiral symmetry restored as we go from low to high energies or, equivalently, from low to high temperatures? A standard approach to the chiral phase transition is to invoke universality arguments [8], namely, it is assumed that the transition is controlled by symmetries rather than by the dynamical details of QCD. The chiral phase transition is then studied by looking at the most general renormalizable Ginzburg-Landau Lagrang...