Mild hypothermia is assumed to protect against secondary brain injury. However, the accuracy of brain temperature estimation remains debatable if direct measurement in the target area is to be avoided or is impossible. Furthermore, intracerebral temperature gradients exist, especially under intraoperative conditions. We aimed to establish how brain surface temperatures (TBrain) relate to temperatures taken at standard sites in posterior fossa surgery. Ten patients undergoing cerebellopontine angle tumor removal were monitored for TBrain, esophageal temperature (TEso), bladder temperature (TBlad), ipsi- and contralateral tympanic membrane (TTymp-I, TTymp-c), and scalp temperatures (TScalp). During monitoring, TEso increased from 35.3 ± 0.2°C to 36.0 ± 0.3°C. After dura opening, TBrain was −0.14 ± 0.1°C below TEso. At the end of tumor removal, this difference increased to −0.43 ± 0.31°C (P < 0.05). TTymp-c was −0.29 ± 0.18°C below TBrain at dura opening. TTymp-c reflected the behavior of TEso adequately (r = 0.938), however, with a mean difference of −0.39 ± 0.04°C. In contrast, TTymp-I readings closely followed temperature changes in the area of surgery. TBlad reflected TEso except in periods of rapid temperature changes. In posterior fossa (PF) surgery, local TBrain is most accurately reflected by TEso. For clinical use TBlad and TTymp-c are also sufficient to assess brain surface temperature in the PF. Intraoperative surface cooling of the brainstem is less than the previously described cooling rate of exposed cerebral cortex.