The present paper deals with the problem of self-sustained hydrodynamic oscillations of a two-phase fluid traveling in a single heated channel. The aim of the work described has been to establish a physically adequate analytical model of this often discussed phenomenon that is simple enough to provide some understanding of the driving mechanisms and to permit immediate conclusions as to the bearing on system stability of several important design and operational parameters. For this reason, it was decided to avoid conventional transfer-function description, which could have provided somewhat greater accuracy. It has been shown that a pure density effect is capable of explaining the oscillatory behavior under the boundary condition of a constant pressure drop over the system. An instructive stability plot that permits a quick survey over the stability properties of a given system is introduced. However, it has also been shown that the integrated or point model is not adequate for all cases. Hence, a modification, called the “long channel correction,” that successfully substitutes a spatial analysis whenever necessary is introduced. Stability plots are obtained that clearly display the difference between the predictions of the model without the long channel correction (the simple model) and the model with the long channel correction. Comparisons have been made to experimental data as well as to the predictions of a stability code with a detailed spatial description. Good agreement has been demonstrated for the model with the long channel correction. The existence of a crucial boiling length or point of minimum stability is confirmed by the corrected model but not by the simple model. Moreover, it appears that the point of minimum stability is closely related to the point where the long channel correction becomes important. In other words, as subcooling is decreased, the increasing importance of the time delays in the system gives rise to an inversion of the stability trend and, at the same time, the point-model description becomes insufficient. The present model is expected to find application in cases where a quick survey of the stability trends of a group of systems is more important than accurate predictions for one particular situation.