Background: The commonly used published Frascati criteria likely overestimate HIV-associated neurocognitive disorder (HAND) prevalence. Modified criteria proposed by Gisslen attempt to reduce sensitivity while retaining specificity. Multivariate normative comparison (MNC), specifically designed to control for false-positive rate while retaining sensitivity, is a novel and potentially more accurate technique for evaluating cognitive impairment (CI). Methodology: 103 HIV-infected men with suppressed viraemia on cART for ≥12 months and 75 uninfected male controls (comparable regarding age, ethnicity, IQ and education), all aged ≥45 and participating in the AGEhIV cohort study, underwent neuropsychological assessment (covering fluency, attention, information processing speed, executive function, memory and fine motor function). Frascati and Gisslen criteria were applied to detect HAND. Next, MNC was used to compare all test scores of each HIV-positive individual against the distribution of test scores of the total control group. MNC uses Hotelling’s T2 to statistically compare multiple characteristics of an individual against the distributions of the same characteristics in a control group taking the covariance between all characteristics into account. The false positive rate (alpha, which is selectable at any percentage) was set at 5% one-tailed, providing a specificity of at least 95%. Results: HIV-positive men as a group performed worse compared to uninfected controls on all domains tested, although differences were not statistically significant. Applying Frascati criteria, HAND was present in 49 HIV-positive (48%) but also in 31 uninfected men (41%) (p=0.41), and by Gisslen’s criteria in 15 HIVpositive (15%) and 7 uninfected men (9%) (p=0.30). Using MNC, CI was detected in 32 HIV-positive men (31%) (Table). Conclusions: Cognitive impairment by Frascati criteria was highly prevalent in HIV-positive but nearly equally so in uninfected men, confirming low specificity of this method. Although HAND prevalence was reduced applying Gisslen’s criteria, it remained nearly as prevalent among uninfected controls, indicating improved sensitivity but continued low specificity. Applying MNC with its low false-positive rate provided a more accurate estimate of HIVassociated cognitive impairment of approximately 30 percent in HIV-infected men with suppressed viraemia on cART.