In a series of 40 patients undergoing an awake craniotomy for the removal of a glioma of the dominant hemisphere temporal lobe, cortical stimulation mapping was used to localize essential language sites. These sites were localized to distinct temporal lobe sectors and compared with 83 patients without tumors who had undergone language mapping for the treatment of intractable epilepsy. In patients with and without temporal lobe gliomas, the superior temporal gyrus contained significantly more language sites than the middle temporal gyrus. Both patient populations also had language sites anterior to the central sulcus in the superior temporal gyrus (12-16%). The patients without tumors had significantly more language sites in the superior temporal gyrus, compared with the superior temporal gyrus of patients with temporal lobe tumors. Multiple variables were studied for their effect on preoperative and postoperative language deficits and included age, sex, number of language sites, histology, size of the tumor, and the distance of tumor resection margins from the nearest language site. The distance of the resection margin from the nearest language site was the most important variable in determining the improvement in preoperative language deficits, the duration of postoperative language deficits, and whether the postoperative language deficits were permanent. If the distance of the resection margin from the nearest language site was > 1 cm, significantly fewer permanent language deficits occurred. Cortical stimulation mapping for the identification of essential language sites in patients with gliomas of the dominant hemisphere temporal lobe will maximize the extent of tumor resection and minimize permanent language deficits.