Background: In patients with intracerebral haemorrhage (ICH), early haemorrhage expansion affects clinical outcome. Haemostatic treatment reduces haematoma expansion, but fails to improve clinical outcomes in many patients. Proper selection of patients at high risk for haematoma expansion seems crucial to improve outcomes. In this study, we aimed to prospectively validate the CT-angiography (CTA) spot sign for prediction of haematoma expansion. Methods: PREDICT (predicting haematoma growth and outcome in intracerebral haemorrhage using contrast bolus CT) was a multicentre prospective observational cohort study. We recruited patients aged 18 years or older, with ICH smaller than 100 mL, and presenting at less than 6 h from symptom onset. Using two independent core laboratories, one neuroradiologist determined CTA spot-sign status, whereas another neurologist masked for clinical outcomes and imaging measured haematoma volumes by computerised planimetry. The primary outcome was haematoma expansion defined as absolute growth greater than 6 mL or a relative growth of more than 33% from initial CT to follow-up CT. We reported data using standard descriptive statistics stratified by the CTA spot sign. Mortality was assessed with Kaplan-Meier survival analysis. Findings: We enrolled 268 patients. Median time from symptom onset to baseline CT was 135 min (range 22-470), and time from onset to CTA was 159 min (32-475). 81 (30%) patients were spot-sign positive. The primary analysis included 228 patients, who had a follow-up CT before surgery or death. Median baseline ICH volume was 19·9 mL (1·5-80·9) in spot-sign-positive patients versus 10·0 mL (0·1-102·7) in spot-sign negative patients (p<0·001). Median ICH expansion was 8·6 mL (-9·3 to 121·7) for spot-sign positive patients and 0·4 mL (-11·7 to 98·3) for spot-negative patients (p<0·001). In those with haematoma expansion, the positive predictive value for the spot sign was 73%; the negative predictive value was 84%, sensitivity was 63%, and specificity was 90%. Median 3-month modified Rankin Scale (mRS) was 5 in CTA spot-sign-positive patients, and 3 in spot-sign-negative patients (p<0·001). Mortality at 3 months was 43·4% (23 of 53) in CTA spot-sign positive versus 19·6% (31 of 158) in CTA spot-sign-negative patients (HR 2·4, 95% CI 1·4-4·0, p=0·002). Interpretation: These findings confirm previous single-centre studies showing that the CTA spot sign is a predictor of haematoma expansion. The spot sign is recommended as an entry criterion for future trials of haemostatic therapy in patients with acute ICH. Funding: Canadian Stroke Consortium and NovoNordisk Canada.