Thermal ablation therapy can potentially offer several advantages over traditional surgery in the treatment of both benign and malignant lesions. Compared to surgery, thermal ablation can reduce the length of hospital stays and significantly reduce overall treatment cost, and may prove superior for treating deep-seated or otherwise difficult-to-reach tumors. Furthermore, the minimally-invasive nature of most thermal ablation treatments may reduce trauma and pain for patients, a particularly attractive advantage in the context of end-of-life palliative care. Heat may be delivered with lasers, RF antennas, or ultrasound transducers, and magnetic resonance (MR) thermometry has been used to monitor thermal ablations. The imaging data must help confirm whether heat is being delivered accurately at the targeted location, help determine when a lethal thermal dose has been reached at the target, and make sure that non-targeted locations are not inadvertently damaged. Demands on the imaging method are especially challenging, as good resolution is required along spatial, temporal and temperature dimensions simultaneously. While MR thermometry has been successfully used during thermal ablations of lesions seated in organs that are fairly easily immobilized, such as the brain and uterus, MR monitoring in more mobile organs such as the liver still remains a significant challenge. Especially in the context of focused ultrasound ablation, where there may not be any device or needle inserted into the body, the task of detecting and resolving motion may fall back entirely onto the imaging method. MR monitoring must then provide, in addition to good thermometry measurements, a temporal resolution sufficient to resolve the motion that occurs in free-breathing patients during treatment. The proposed work is aimed at developing an acquisition, reconstruction and display package for the MR monitoring of thermal ablations in mobile organs such as the liver, capable of providing limited 3D coverage with a temporal resolution of about 1/2 second or better. A novel pulse sequence design and reconstruction strategy is proposed, which is specifically targeted toward tackling the motion problem in MR thermometry imaging. The proposed approach can provide distortion-free 3D anatomical data with good signal-to-noise ratio as well as flexible contrast for motion tracking, while also simultaneously providing accurate temperature measurements. By exploiting the flexibility in the obtained contrast, one may preferentially highlight given internal features such as the vascular bed, thus improving the ability of registration algorithms to detect and track motion.