Three-dimensional positron emission tomography (PET) simulation procedures based upon a digital brain phantom and tomographic detector response functions have been performed. The recovery of PET signals from anatomical brain structures and subsequent error propagation into both static and kinetic data analysis stratagies have been investigated for the measurement of cerebral blood flow, metabolism, and neurotransmission systems. Errors created by the limitations of the PET tomography in human neurologic studies for both static and dynamic data-acquisition schemes are evaluated based upon the simulation techniques. It is shown that careful attention to the definition of the 3D distribution of radionuclides and the characteristics of PET tomographic systems enable the production of reliable results. Applications of this tomographic simulation approach will include the evaluation of study feasibility and the optimization of data analysis strategies.