In two measures theories (TMT), in addition to the Riemannian measure of integration, being the square root of the determinant of the metric, we introduce a metric-independent density Phi in four dimensions defined in terms of scalars varphi _a by Phi =varepsilon ^{mu nu rho sigma } varepsilon _{abcd} (partial _{mu }varphi _a)(partial _{nu }varphi _b) (partial _{rho }varphi _c) (partial _{sigma }varphi _d). With the help of a dilaton field phi we construct theories that are globally scale invariant. In particular, by introducing couplings of the dilaton phi to the Gauss–Bonnet (GB) topological density , {sqrt{-g}} , phi left( R_{mu nu rho sigma }^2 - 4 R_{mu nu }^2 + R^2 right) , we obtain a theory that is scale invariant up to a total divergence. Integration of the varphi _a field equation leads to an integration constant that breaks the global scale symmetry. We discuss the stabilizing effects of the coupling of the dilaton to the GB-topological density on the vacua with a very small cosmological constant and the resolution of the ‘TMT Vacuum-Manifold Problem’ which exists in the zero cosmological-constant vacuum limit. This problem generically arises from an effective potential that is a perfect square, and it gives rise to a vacuum manifold instead of a unique vacuum solution in the presence of many different scalars, like the dilaton, the Higgs, etc. In the non-zero cosmological-constant case this problem disappears. Furthermore, the GB coupling to the dilaton eliminates flat directions in the effective potential, and it totally lifts the vacuum-manifold degeneracy.