The experimental tests of the strong interactions have not provided any hint of CP violation in QCD-mediated processes. The prototypical CP-odd observable is the electric dipole moment of the neutron, for which there are stringent bounds. However, the QCD Lagrangian admits a CP-odd topological term proportional to the so-called theta angle, which weighs the contributions to the partition function from different topological sectors. The generic expectation is that such a CP-odd interaction should lead to a nonzero neutron dipole moment, unless there is a severe tuning of the theta angle against the phases of the quark masses. This need of tuning to explain experimental observations constitutes the so-called strong CP problem. In this talk, we challenge the conventional view by showing that taking the spacetime volume to infinity in a consistent manner leads to the theta angle dropping out of correlation functions, so that it becomes unobservable and the CP symmetry is automatically preserved without the need of tuning. We arrive at this result by either using instanton computations or by relying on general arguments based on the cluster decomposition principle and the index theorem.