Experiments by Chrétien and co-workers suggest that mitochondria maintain a temperature that can be hotter than their surroundings by as much as 10 °C. Theor. considerations by others suggest temperature differences that are smaller by several orders of magnitudes (10-5 °C). Nevertheless, the exptl. observations stand unfalsified, begging for an explanation of this million-fold disagreement. We have, hence, a "mitochondrion paradox" at hand. The paradox is resolved by an examination of the thermal conductance of ATP synthase and of the mechanisms of heat production and dissipation and their time-scales. Our results suggest that every proton that passes from the inter-membrane gap to the matrix sparks a picosecond temperature-difference spike of the order of magnitude measured by Chétien et al. When these ps spikes are averaged over the time between successive protons translocations through ATP synthase the theor. value of 10-5 °C is recovered. The paradox is, hence, resolved by the understanding that Chétien et al. are measuring a temporal and ensemble statistical average of the fluorescence spectra of the mito-thermo yellow (MTY) dye they use as a mol. thermometer. Meanwhile, the theoreticians are also correct in that such a huge temperature gradient is impossible to maintain as a steady-state rather than as ps spikes. The image that emerges suggests a dynamic inner mitochondrial membrane flanked by temperature differences fluctuating both in time and along the membrane's surface, with localized "hot" and "cold" regions and ultrashort temperature spikes.