OBJECTIVE: To describe the relationship between intraventricular hemorrhage (IVH) expansion and long-term outcome and to use this relationship to select and validate clinically relevant thresholds of IVH expansion in 2 separate intracerebral hemorrhage (ICH) populations. METHODS: We used fractional polynomial analysis to test linear and nonlinear models of 24-hour IVH volume change and clinical outcome with data from the Predicting Hematoma Growth and Outcome in Intracerebral Hemorrhage Using Contrast Bolus CT (PREDICT)-ICH study. The primary outcome was poor clinical outcome (modified Rankin Scale [mRS] score 4-6) at 90 days. We derived dichotomous thresholds from the selected model and calculated diagnostic accuracy measures. We validated all thresholds in an independent single-center ICH cohort (Massachusetts General Hospital). RESULTS: Of the 256 patients from PREDICT, 127 (49.6%) had an mRS score of 4 to 6. Twenty-four-hour IVH volume change and poor outcome fit a nonlinear relationship, in which minimal increases in IVH were associated with a high probability of an mRS score of 4 to 6. IVH expansion ≥1 mL (n = 53, sensitivity 33%, specificity 92%, adjusted odds ratio [aOR] 2.68, 95% confidence interval [CI] 1.11-6.46) and development of any new IVH (n = 74, sensitivity 43%, specificity 85%, aOR 2.53, 95% CI 1.22-5.26) strongly predicted poor outcome at 90 days. The dichotomous thresholds reproduced well in a validation cohort of 169 patients. CONCLUSION: IVH expansion as small as 1 mL or any new IVH is strongly predictive of poor outcome. These findings may assist clinicians with bedside prognostication and could be incorporated into definitions of hematoma expansion to inform future ICH treatment trials.