Hypoxic tumors are associated with poor patient prognosis because hypoxia leads to angiogenesis, metastasis, aggressiveness of the cancer, and resistance to chemotherapy and radiation therapy. This dissertation explores imaging modalities to accurately measure and locate tumor hypoxia to improve radiation therapy. We used three tumor murine models of squamous cell carcinomas (SCC7), mammary adenocarcinomas (MCa-4), and fibrosarcomas (FSa) for electron paramagnetic resonance oxygen imaging (EPROI), [18F]-Fluoromisonidazole positron emission tomography (FMISO PET), dynamic contrast-enhanced magnetic resonance imaging (DCE MRI), and histological imaging. The theme of this dissertation is using EPROI as an in vivo validation of absolute pO2, which is traditionally an invasive and discretely measured task. For now, EPROI is generally a preclinical imaging tool that is rarely available in the clinic. However, we can compare measurements from EPROI to the more clinically available modalities, like FMISO PET and DCE MRI, to develop tools to improve their accuracy. This was accomplished in four parts. First, we used EPROI to demonstrate the effectiveness of delivering a radiation boost to more resistant hypoxic tumor regions, while minimizing radiation dose to oxygenated tumor regions and surrounding healthy tissue. Local tumor control probability improved by at least a factor of two when comparing hypoxic versus oxygenated boost treatment groups. Theseexperiments in oxygen-guided radiation therapy show promise in minimizing dose while improving radiation therapy outcomes. While EPROI has a clear threshold to define hypoxia in vivo (pO2≤10mmHg), there is presently no unifying threshold to define hypoxia with FMISO PET. Here we identified optimal FMISO uptake thresholds to define hypoxia with a custom-built hybrid PET/EPR machine for near-simultaneous hypoxia imaging, using EPROI as ground truth to define tumor hypoxia. The optimal thresholds varied by tumor type, and on average had a 68-73% similarity between hypoxic volumes defined by FMISO PET and EPROI. DCE MRI identified features of tumor vasculature and extracellular-extravascular space that may pinpoint where and why FMISO PET was not as accurate as EPROI in locating tumor hypoxia. EPROI determined where FMISO PET correctly classified or misclassified voxels as normoxic or hypoxic. Additionally, histological images of axial tumor slices stained with H&E validated tumor boundaries and necrosis, and IHC stains of the hypoxia inducible factor 1α (HIF-1α) and vasculature with CD31 were compared to registered in vivo slices. Tying all modalities together, methods of modeling and correcting FMISO PET with pO2 and DCE MRI were evaluated. A logistic model was implemented in a correction algorithm that combines FMISO PET with optimally weighted DCE MRI parametric images to improve the accuracy of hypoxia location. This work sets up future experiments that may use corrected FMISO PET images to locate tumor hypoxia for oxygen image-guided radiation therapy originally done with EPROI. The presented research is a step toward improving radiation therapy methods and outcomes for patients with hypoxic tumors. Throughout, we demonstrate tumor-type dependence of the accuracy of FMISO PET and highlight the effectiveness of oxygen-guided radiation therapy in improving local tumor control.