We investigate the use of Cu1-x Zn x Fe2O4 ferrites (0.60 < x < 0.76) as potential sensors for magnetic- resonance-imaging thermometry. Samples are prepared by a standard ceramic technique. Their structural and magnetic properties are determined using x-ray diffraction, scanning electron microscopy, super-conducting quantum-interference device magnetometry, and Mossbauer and 3-T nuclear-magnetic-resonance spectroscopies. We use the mass magnetization of powdered ferrites and transverse relaxivity r*2 of water protons in Ringer's-solution-based agar gels with embedded micron-sized particles to determine the best composition for magnetic-resonance-imaging (MRI) temperature sensors in the (280-323)-K range. A preclinical 3-T MRI scanner is employed to acquire T*2 weighted temperature-dependent images. The brightness of the MRI images is cross-correlated with the temperature of the phantoms, which allows for a temperature determination with approximately 1°C accuracy. We determine that the composition of 0.65 < x < 0.70 is the most suitable for MRI thermometry near human body temperature.