Lead halide perovskites with mixtures of monovalent cations have attracted wide attention due to the possibility of preferentially stabilizing the perovskite phase with respect to photovoltaically less suitable competing phases. Here, we present a theoretical analysis and interpretation of the phase stability of binary (CH6N3)(x)[HC(NH2)(2)]((1-x))PbI3= GUA(x)FA((1-x))PbI(3)and ternary Cs(y)GUA(x)FA((1-y-x))PbI(3)mixtures. We first estimate if such mixtures are stable and if they lead to a stabilization of the perovskite phase based on static Density Functional Theory (DFT) calculations. In order to investigate the finite temperature stability of the phases, we also employ first-principles molecular dynamics (MD) simulations. It turns out that in contrast to the FA(+)-rich case of FA/Cs mixtures, although mixing of FA/GUA is possible, it is not sufficient to stabilize the perovskite phase at room temperature. In contrast, stable ternary mixtures that contain 17% of Cs(+)can be formed that lead to a preferential stabilization of the perovskite phase. In such a way, the enthalpic destabilization due to the introduction of a too large/too small cation that lies outside the Goldschmidt tolerance range can be (partially) compensated through the introduction of a third cation with complementary size. This allows to suggest a new design principle for the preparation of stable perovskite structures at room temperature with cations that lie outside the Goldschmidt range through mixtures with size-complementary cations in such a way that the effective average cation radius of the mixture lies within the stability range.