Chalcogenide glass is being investigated by several companies as the basis for a scalable and embeddable nonvolatile phase-change memory technology. One phase is a high-resistance amorphous phase that is obtained by melting a small volume of glass using ohmic heating, and then quenching it. The second phase is a low-resistance crystalline phase that is obtained by heating the glass to just below the melting point to promote recrystallization. This paper describes two models for such a cell. The first is a very simple single-element, lumped model that exhibits correct phase transition behavior, but is unrealistic in its sensitivity to the heating current pulses. The second, multiple-element model is able to more realistically represent cell heating and cooling behavior, and appears to be the more suitable basis for an electrical simulation model.