More and more researchers and clinicians are looking to molecular sensing to predict how cells will behave, seeking the answers to questions like will these tumor cells become malignant? or how will these cells respond to chemotherapy? Optical methods are attractive for answering these questions because optical radiation is safer and less expensive than alternative methods, such as CT which uses X-ray radiation, PET/SPECT which use gamma radiation, or MRI which is expensive and only available in a hospital setting. In this dissertation, three distinct optical methods are explored to detect at the molecular level: optical coherence tomography OCT), laser-induced fluorescence LIF), and optical polarimetry. OCT has the capability to simultaneously capture anatomical information as well as molecular information using targeted contrast agents such as gold nanoshells. LIF is less useful for capturing anatomical information, but it can achieve significantly better molecular sensitivity with the use of targeted fluorescent dyes. Optical polarimetry has potential to detect the concentration of helical molecules, such as glucose. All of these methods are noninvasive or minimally invasive. The work is organized into four specific aims. The first is the design and implementation of a fast, high resolution, endoscopic OCT system to facilitate minimally invasive mouse colon imaging. The second aim is to demonstrate the utility of this system for automatically identifying tumor lesions based on tissue microstructure. The third is to demonstrate the use of contrast agents to detect molecular expression using OCT and LIF. The last aim is to demonstrate a new method based on optical polarimetry for noninvasive glucose sensing.