Accurate field-scale simulations of foam enhanced oil recovery are challenging, due to the sharp transition between gas and foam. Hence, unpredictable numerical and physical behavior is often observed, casting doubt on the validity of the simulation results. In this paper, a thorough stability analysis of the foam model is presented to validate the simulation results. We study the effect of a strongly non-monotonous total mobility function arising from foam models on the stability characteristics of the flow. To this end, we apply the linear stability analysis to nearly discontinuous relative permeability functions and compare the results with those of highly accurate numerical simulations. In addition, we present a qualitative analysis of the effect of different reservoir and fluid properties on the foam fingering behavior. In particular, we consider the effect of heterogeneity of the reservoir, injection rates, and foam quality. Relative permeability functions play an important role in the onset of fingering behavior of the injected fluid. Hence, we can deduce that stability properties are highly dependent on the non-linearity of the foam transition. The foam-water interface is governed by a very small total mobility ratio, implying a stable front. The transition between gas and foam, however, exhibits a huge total mobility ratio, leading to instabilities in the form of viscous fingering. This implies that there is an unstable pattern behind the front. We deduce that instabilities are able to grow behind the front but are later absorbed by the expanding wave. Moreover, the stability analysis, validated by numerical simulations, provides valuable insights about the important scales and wavelengths of the foam model. In this way, we remove the ambiguity regarding the effect of grid resolution on the convergence of the solutions. This insight forms an essential step toward the design of a suitable computational solver that captures all the appropriate scales, while retaining computational efficiency.