Three hydrocarbon phases can co-exist at equilibrium at relatively low temperatures in many CO 2 floods. Formation of an aqueous phase in contact with hydrocarbon phases is inevitable in almost all recovery processes, because of the permanent presence of water in the reservoirs either as injection fluid or as initial formation water. Successful modeling of CO 2 flooding requires accounting for the presence of four phases. However, as the number of phases increase, flash calculations become more difficult and time-consuming. A possible approach to reduce the computational time of the phase equilibrium calculations is to use reduced methods. This paper presents a general strategy to model the behavior of CO 2/hydrocarbon/water systems where four equilibrium phases occur using a reduced flash approach. The speedup obtained by a reduced flash algorithm compared to the conventional flash approach is demonstrated for a different number of components and phases. The results show a significant speedup in the Jacobian matrix construction and in Newton-Raphson iterations using the reduced method when four phases are present. The computational advantage of the reduced method increases rapidly with the number of phases and components. The developed four-phase reduced flash algorithm is used to investigate the effect of introducing water on the phase behavior of two West Texas oil/CO 2 mixtures. The results show significant changes in the phase splits and saturation pressures by adding water to these CO 2/hydrocarbon systems.