Research findings have suggested a need for modifications to the original nomenclature for human immunodeficiency virus (HIV)-associated neurocognitive disorders issued in 1991 by the American Academy of Neurology (AAN). The HIV Neurobehavioral Research Center (HNRC) proposed a diagnostic scheme that departs from the AAN 1991 criteria primarily in the inclusion of an asymptomatic neurocognitive impairment (ANI) category that relies on cognitive disturbances as a necessary criterion for diagnosis, without requiring declines in daily functioning, motor, or other behavioral abnormalities. In order to test the predictive validity of these two nomenclatures, the authors compared the correspondence between antemortem neurocognitive diagnoses resulting from AAN and HNRC criteria to a neuropathological diagnosis of HIV encephalitis (HIVE) made at autopsy. Agreement between the two sets of definitional criteria was 79% regarding the classification of cases as either neurocognitively normal or impaired, and 54% with regard to specific neurocognitive diagnoses. When pathological evidence of HIVE was considered as the external indicator of HIV-related brain involvement, 64% of cases were correctly classified by AAN criteria, compared to 72% by HNRC criteria. HNRC criteria had better positive predictive power (95% versus 88%), sensitivity (67% versus 56%), and specificity (92% versus 83%). Three cases with HIVE and were correctly identified by HNRC criteria for ANI but called normal by AAN criteria, supporting inclusion of an asymptomatic neurocognitive condition. The modifications to the AAN 1991 criteria proposed by the HNRC and others in the field have served as a point of departure for a recently revised consensus nomenclature.