Introduction: Hounsfield units (HU) have been proposed as a prognostic marker of hematoma expansion in patients with intracerebral hemorrhage (ICH). The aim of this study was to determine the optimal HU value for predicting hematoma expansion on noncontrast computed tomography (NCCT) in patients diagnosed with ICH. Material and Methods: Prospective cohort study of patients with ICH. Baseline cranial NCCT was performed within 8 hours of symptom onset and an NCCT follow-up within 72 hours of the baseline examination. HU and the shape and density of the hematoma were assessed at baseline NCCT. The optimal HU cut-off value for predicting hematoma expansion was determined by ROC analysis. Results: Fifty patients diagnosed with ICH were included. There were 28 (56.0%) men and 22 (44.0%) women. Their mean age was 57.8 ± 16.97 years. Hematoma expansion was present in 14 (28.0%) cases. At baseline NCCT, the mean HU was significantly lower in patients with expansion (61.1 ± 2.8) than in patients without hematoma expansion (69.2 ± 5.5) (p < 0.001). A value equal to or less than 64.2 HU on baseline NCCT was defined as the optimal cut-off point for predicting hematoma expansion. Sensitivity was 78.6%, specificity 86.1%, and the AUC was 0.914 (95% CI, 0.834-0.993) (p < 0.001). Conclusion: In this study, for the first time, a value equal to or less than 64.2 HU by ROC analysis on baseline NCCT was defined as the optimal cut-off point for predicting hematoma expansion in patients with ICH.