The small magnitude of (T, -T )/T"where T, is the nematic-isotropic phase transition temperature and T denotes the virtual transition temperature, has been a long-standing puzzle in the physics of liquid crystals. We show that by extending the mean field theory to include the isotropic, densitydependent component of the molecular interaction the magnitudes of both (T, -T*)/T, and the density change at the transition automatically become in accord with the experimental values. In addition, the theory yields a value of dT, /dp, where p denotes pressure, that is on the same order as the experiment. [4], but the incorporation of that extra part also materially affects the character of the N-I transition.In particular, the predicted magnitude of (T, -T*)/T, is lowered to the experimental range, while at the same time the theory also yields the right magnitudes both for the density variation across the N-I transition as well as for the variation of T, with pressure.