A theoretical and numerical analysis of spectral self-interference microscopy (SSM) is presented with the goal of expanding the realm of SSM applications. In particular, this work is intended to enable SSM imaging in low-signal applications such as single-molecule studies. A comprehensive electromagnetic model for SSM is presented, allowing arbitrary forms of the excitation field, detection optics, and tensor sample response. An evanescently excited SSM system, analogous to total internal reflection microscopy, is proposed and investigated through Monte Carlo simulations. Nanometer-scale axial localization for single-emitter objects is demonstrated, even in low-signal environments. The capabilities of SSM in imaging more general objects are also considered--specifically, imaging arbitrary fluorophore distributions and two-emitter objects. A data-processing method is presented that makes SSM robust to noise and uncertainties in the detected spectral envelope.