<i>Background:</i> The Hemorrhage after Thrombolysis (HAT) score has recently been introduced as a practical scale for risk stratification of intracranial hemorrhage (ICH) in patients receiving intravenous tissue plasminogen activator (tPA). We aimed to externally validate and evaluate the predictive ability of the HAT score in patients with proximal arterial occlusions (PAO) enrolled into randomized clinical trials of sonothrombolysis. <i>Methods:</i> The HAT score (range 0, minimum risk, to 5, maximum risk) was retrospectively calculated for each patient using clinical trial data (baseline NIHSS, extent of hypodensity on CT, history of diabetes mellitus and serum glucose). Symptomatic ICH (sICH) was defined as imaging evidence of ICH with clinical worsening (NIHSS ≧4) within 72 h from stroke onset. The predictive ability of the HAT score for sICH and any ICH (both asymptomatic and symptomatic) was calculated using c statistics. <i>Results:</i> A total of 161 tPA-treated patients (mean age 68 ± 13 years, 58% men, median NIHSS 16, interquartile range 9) with PAO were randomized in TUCSON (n = 35) and CLOTBUST (n = 126). sICH occurred in 9 (5.6%) cases, and 6 had asymptomatic ICH. The rates of sICH for the corresponding HAT scores were: HAT 0–1: 3%; 2: 9%; 3: 14%; 4–5: 14%. The risk of sICH (c statistic 0.72, 95% CI: 0.58–0.86; p = 0.027) and any ICH (c statistic 0.70, 95% CI: 0.58–0.82; p = 0.011) increased with higher HAT scores. Higher HAT scores were also associated with higher likelihood of persisting occlusion (c statistic 0.63, 95% CI: 0.54–0.72; p = 0.004). <i>Conclusions:</i> The HAT score has reasonable external validity for predicting the risk of sICH following intravenous thrombolysis in patients with PAO. Moreover, higher HAT scores appear to be associated with higher likelihood of persisting occlusion in tPA-treated patients.