The induction of temperature elevation by focused ultrasound is a noninvasive surgical technique for destroying tissue. This technique has been used clinically in soft tissues such as liver, prostate and breast. It has long been desired to extend this technique to noninvasive treatment of brain tumors. Although the skull was once thought to be an unsurpassable barrier to focused ultrasound treatment, it has been shown that the distortion caused by the skull can be corrected to produce a useful intracranial focus. However, the attenuation experienced by the ultrasound in passing through cranial bone is large, and consequently the skull is subject to the deposition of acoustic energy as heat. The nature and extent of this heating process has been difficult to characterize empirically. It is practically difficult to implant a sufficient number of thermocouples to obtain detailed temperature data directly, and bone is an unsuitable medium in which to perform noninvasive thermometry using proton chemical shift magnetic resonance imaging. Furthermore, skull specimens used experimentally lack active blood perfusion of the skull and the overlying scalp. This paper describes the use of large-scale acoustic and thermal simulations to calculate the distribution of temperature within the skull and brain that can be expected to occur during therapeutically useful focused ultrasound sonications of the brain. The results demonstrate that standing waves may be formed within the skull during transcranial sonication leading to nonuniform skull heating. However, the results also show that these effects can be sufficiently controlled to allow therapeutic ultrasound to be focused in the cranial base region of the brain without causing thermal damage to the scalp, skull or outer surface of the brain.