Both X-ray computed tomography (CT) and digital breast tomosynthesis (DBT) have become important clinical tools for the non-invasive assessment and diagnosis of disease in the breast. The continuing development and refinement of image reconstruction methods has led to significant improvements in image quality in both of these modalities. In particular, sparsity exploiting iterative image reconstruction (IIR) methods have demonstrated potential for yielding images of improved quality from the high-noise and limited angular data encountered in breast CT and DBT, respectively; however, realization of these improvements typically requires manual tuning of numerous parameters, often in a case-by-case fashion. Even in the absence of complications introduced by the incorporation of sparsity-exploiting regularizers, it is often difficult to establish a direct link between the parameters involved in specifying an IIR algorithm and the quality of the resulting reconstruction. This thesis aims to address this issue in the context of breast CT and DBT. As a step towards this goal, we simplify the problem by considering IIR without sparsity exploiting regularizers. The primary focus of this thesis is the characterization of parameter trends for such IIR algorithms in breast CT and DBT.