The fact that common thermodynamic conditions are valid for all known types of critical phases (liquid-liquid, liquid-gas, and "gas-gas") suggests that a common principle for the interpretation of material phase instability from a molecular point of view must exist. In this paper we show that the principle of competition between "statistical mixing" (i. e. random mixing) and "chemical mixing" (i. e. mixing effected under the influence of chemical interactions) can give this common inter pretation. If the equilibrium states resulting from both types of mixing are sufficiently different, phase separation occurs. We refer to our earlier papers (since 1972) in which we have applied this principle to describe liquid-liquid phase equilibria by "chemical" models, using the equilibrium constants of exchange equilibria between nearest-neighbour complexes as a measure of "chemical" mixing. In this paper we show that the well-known reduced gas-liquid coexistence curve, T/T c =f(q/q c ), can accurately be fitted by a very simple "mixture" model of molecules A with "vacan cies", provided that the contributions of both statistical and chemical mixing are incorporated into the formula for G E . From a discussion of the application to "gas-gas" phase equilibria in the hyper critical region it results that the weight factor r, by which the contribution of statistical mixing enters into G E , must depend on the density of the gas mixture. Phase separation can only occur if, by increasing pressure, the contributions to G E of statistical and chemical mixing have reached the same order of magnitude. From an attempt to apply the same principle to solid-liquid equilibria it is shown under which external conditions a critical point for this type of phase transition can be expected.