Flat-panel-based cone-beam CT (CBCT) has become an important tool for Image-guided radiation therapy (IGRT). However, repeated use of CBCT imaging can result in a high accumulated radiation dose, which poses a health concern to patients. Current CBCT for IGRT also suffers from poor image quality. Thus it is important to reduce patient dose, and improve image quality of current CBCT used in IGRT. In recent years, a great deal of effort has been devoted to the development of optimization-based (i.e. iterative) image reconstruction algorithms for the purpose of reducing CBCT imaging dose or improving CBCT image quality. It has been shown that iterative reconstruction techniques may yield images of improved quality over standard analytic-based algorithms from both low-dose diagnostic CBCT data and sparse-view flat-panel-based CBCT data. However, CBCT image quality is related with both the exposure level per view and number of views at which projections are collected. It is important to explore the trade-off between the view number and exposure level for optimization-based algorithms for the purpose of both dose reduction and image quality improvement purposes. In this work, we performed a preliminary study of optimizing the CBCT imaging quality for IGRT with optimization-based reconstruction algorithms. The preliminary results show that the image quality of ASD-POCS is relatively stable across a wide range of views and exposure levels and seems to indicate that view sampling around one view per degree is more optimal than a higher sampling density with the same total exposure.