Index terms: temperature mapping; proton frequency shift; RF thermal ablation; interventional MR; medical instrumentation MINIMALLY INVASIVE PERCUTANEOUS thermoablation techniques, such as radio frequency (RF) ablation, have shown promise in the treatment of deep-seated tumors in the human body (1,2). Compared to conventional surgical interventions, these techniques can reduce overall morbidity and shorten patient recovery times, thereby decreasing health-care costs. Temperature sensitivity inherent to MR experiments provides the basis for using magnetic resonance imaging (MRI) to monitor thermosensitive therapies. Several temperature-dependent MR parameters, such as the spin-lattice relaxation time T1 (3-5), the molecular diffusion coefficient (6,7), or the proton frequency shift (PFS) (8 -10) can be exploited for the purpose of MR temperature mapping.Fast T1-weighted sequences have been used to monitor local heating of various organs (5,11,12). However, T1 temperature dependency varies in different tissues and is influenced by thermoregulative processes and metabolic tissue changes. Hence, temperature changes cannot be reliably quantified. Diffusion-weighted imaging is based on the thermal Brownian motion, which is described by the diffusion coefficient. The drawback of this method lies in long data acquisition times. If applied in vivo, it suffers from thermoregulative diffusion changes and from tissue motion. The PFS technique takes advantage of the temperature-dependent phase shift of the MR signal (8 -10). PFS has very small tissue dependency of the chemical shift compared to the T1 relaxation time and relatively short acquisition time compared to the diffusion technique. PFS temperature sensitivity is related to changes of the molecular screening constant. A disadvantage of the PFS technique lies in its sensitivity to motion artifacts and frequency drifts induced by system instabilities. While motion artifacts can be reduced to some extent by breath holding, system imperfections may induce a phase offset, which can be reduced by suitable correction schemes.The availability of 0.5 T open-configuration systems, in conjunction with optical tracking systems for realtime instrument guidance in an MR environment (15,16), allows for easy placement and interactive positional adjustment of thermal therapy applicators in various organs. For clinically effective monitoring, MR temperature maps need to be acquired in near real time with sufficient spatial and contrast resolution. Furthermore, reliable temperature control must be tissue independent and provide quantitative temperature data.The purpose of this work was to validate the PFS technique for quantitative temperature mapping in an