The second part of this two-part paper summarizes work on the micromechanical modeling of polycrystalline shape memory alloys (SMAs). Averaging micromechanics methods based on the self-consistent approximation are used for the modeling of polycrystalline SMA thermomechanical behavior. The predictions of several models are directly compared and correlated with experimental results. Rate independent phenomenological models are then discussed, which are based on characterizing the inelastic fields associated with the phase transformation and transformation induced plasticity by using internal state variables. The resulting evolution equations are integrated using return mapping algorithms. Selected numerical simulations and comparison of the phenomenological models with the micromechanical ones are also presented.