CERAD, a multicenter longitudinal study, has developed standardized instruments for the evaluation of individuals clinically diagnosed as having Alzheimer's disease (AD). The CERAD neuropathology protocol not only establishes levels of certainty for AD diagnosis, but also records information on other conditions occurring concomitantly with or misdiagnosed as AD. The protocol has been widely adopted because of its relative simplicity, adaptability, and reliability. Indeed, the Consensus principles proposed are, for the most part, consistent with CERAD guidelines. The recommendation that diagnosis rest upon both neuritic plaque and neurofibrillary tangle frequency/distribution in the neocortex, however, is worrisome. This change will eliminate or downgrade many cases now diagnosed as AD with concomitant Parkinson's disease changes. Reclassifying such cases at this time, without compelling pathobiological justification, is premature. Instead, I recommend retention or modest modification of the current CERAD protocol, and propose that neuropathology data available on autopsies of over 200 CERAD dementia subjects be used for testing potential modifications of the diagnostic algorithm.