CT was used to document temporal lobe atrophy in 39 patients who were diagnosed clinically as having Alzheimer disease; the results were compared with those from 29 healthy elderly control subjects who were matched for age and education. The diagnosis of Alzheimer disease was made according to clinical criteria consistent with those specified by an NINCDS workshop. These included detailed medical and neuropsychological assessments. Temporal lobe atrophy was assessed by evaluating the temporal horns and sylvian cisterns. Temporal horn measurements greater than 3 mm occurred only in patients with Alzheimer disease while measurements less than or equal to this occurred in both Alzheimer patients and control subjects. Subjective evaluation of the sylvian cistern indicated that 24/29 controls had normal-appearing cisterns while only 5/39 Alzheimer patients had similar findings. In contrast to temporal horns, sylvian cisterns were more sensitive but less specific as discriminators between Alzheimer patients and normal controls.