Positron emission tomography regional CBF (rCBF) studies of cognitive processes have traditionally required 30-60 mCi of H2(15)O per scan and intersubject averaging to achieve statistical significance. However, intersubject anatomical, functional, and disease variability can make such an approach problematic. A new method that produces significant results in single subjects is presented. It is based upon high-sensitivity three-dimensional imaging and a "slow" bolus administration of < 15 mCi of H2(15)O per scan. The method is validated in four normal volunteers using control and auditory-language activation tasks with four scans per condition and statistical parametric mapping analysis. It is demonstrated that the rCBF distribution associated with the cognitive state is detected during the arrival of radiotracer in the brain. This occurs over 30 s and constitutes a critical temporal window during which stimulation should be performed. A 90-s acquisition time is found to produce results of greater significance than a 60-s acquisition time. The implications of the results and the functional neuroanatomical findings are discussed. This method is suitable for the study of individual functional neuroanatomy in many neuropsychological, pharmacologic, and symptom states in normal subjects and in patients with psychiatric and neurologic disorders.