Abstract. We shall present a new class of conservative difference approximations for the steady full potential equation. They are, in general, easier to program than the usual density biasing algorithms, and in fact, differ only slightly from them. We prove rigorously that these new schemes satisfy a new discrete "entropy inequality", which rules out expansion shocks, and that they have sharp, steady, discrete shocks. A key tool in our analysis is the construction of an "entropy inequality" for the full potential equation itself. We conclude by presenting results of some numerical experiments using our new schemes.I. Introduction. The full potential equation is a common model for describing supersonic and subsonic flow close to the speed of sound. The flow is assumed to be that of a perfect gas, and the assumptions of irrotationality and constant entropy are made. The resulting equation is a single nonlinear partial differential equation of second order, which changes type from hyperbolic to elliptic, as the flow goes from supersonic to subsonic. Flows with a supersonic component generally have solutions with shocks, so the conservation form of the equation is important.This formulation, (FP), is one of three conservative formulations used for inviscid transonic flows. The other two are transonic small-disturbance equation, (TSD), and Euler equation, (EU), which is the exact inviscid formulation. The FP formulation is the most efficient of the three in terms of accuracy-to-cost ratio for a wide range of inviscid transonic flow applications for real geometries. TSD is valid for thin wings at free stream Mach numbers near unity, and EU, while the least restrictive, involves the most complicated system of equations.During the last few years, many numerical calculations using FP have been presented, e.g., [19], [14], [17], and [6]. The object of our present investigation is twofold. First, we wish to put the theory of nonlinear difference approximations to FP on a sound theoretical basis, via an "entropy condition", as described below. Second, we introduce a new class of entropy condition satisfying approximations, which are, in general, no more complicated to program than the usual density