The subjective user satisfaction related to viewing a video on a resource constrained mobile device is modeled by using a utility function. The resulting models account for different video content characteristics and varying hardware capabilities of the viewing device. The user satisfaction is divided into two independent components, namely the temporal and the spatial satisfaction. Subjective video evaluation experiments are performed to obtain the models for each component. These components are combined by using a weighted sum. The firame-rate and the bit-rate that maximizes the total utility is then obtained by the optimization algorithms. It is further demonstrated that the utility curves obtained through subjective tests for a limited number of devices can be interpolated to generate utility curves for other devices with different hardware capabilities.