Radiofrequency fields are ubiquitously used in Magnetic Resonance Imaging to excite water protons and form images. A non-uniform tissue temperature increase results from the balance between this power deposition and heat dissipation. Regulations impose that brain temperature should not locally exceed 39°C in normal exam conditions. Yet, because the radiofrequency field-induced temperature rise is so small, it has remained impossible to verify this safety limit in vivo on human subjects. As a result, MR practice has relied instead on electromagnetic simulations and experiments in vitro and on animals. However, anatomical and physiological differences, including anaesthesia, blur the extrapolation to humans. A methodology is introduced to measure in the brain sub-degree temperature rises with MR thermometry, field monitoring and motion compensation. Results show barely any temperature increase on awake humans with a head coil. Potential implications include improvement of human thermal models and revision of safety norms.