We aimed to investigate the value of the hyperdense basilar artery (HBA) sign and of basilar artery (BA) attenuation measurements as predictors of basilar artery occlusion (BAO) on nonenhanced cranial CT (NECT). Forty-one consecutive patients with proven BAO in CT angiography, who had undergone NECT for initial evaluation (30 males, 11 females) were retrospectively included. Another 41 age-matched patients without BAO were included as a control group. The NECT scans of both groups were assessed by three independent blinded readers (staff, fellow, and resident) in a randomized reading order using a standardized semiquantitative questionnaire. Visual BA hyperdensity, including the presence of HBA sign (hyperdensity scores of 4 and 5/5), was assessed, quantitative BA attenuation was measured in a region of interest (ROI), and diagnosis of BAO was made before and after ROI measurements. For statistical analysis, multivariate mixed effects models, likelihood ratio tests, and receiver operating characteristics techniques were applied. HBA sign had a relatively low sensitivity (60.98–65.85%), specificity (70.73–90.24%), and accuracy (65.85–75.61%) for the presence/absence of BAO on NECT. Optimal cut-off points were 40–42 HU (sensitivity, 68.29–78.05%; specificity, 75.61–82.93%; accuracy, 74.39–80.49%). In basilar artery occlusion, quantitative measurement of BA attenuation can slightly improve the diagnostic predictiveness of NECT. However, even with optimal cut-off values, the sensitivity is too low to serve as the sole diagnostic decision-making tool.