Epilepsy outcomes after therapeutic hypothermia for neonates with hypoxic-ischemic encephalopathy are understudied. The authors used multivariable logistic regression to predict epilepsy in neonates after selective head cooling. Sensitivity analyses used magnetic resonance imaging (MRI) and electroencephalogram (EEG) interpretations by different clinicians. Fifty neonates had 2-year follow-up. Nine developed epilepsy. Predictors included pH ≤6.8 on day of birth (adjusted odds ratio [OR] 19 [95% confidence interval (CI) 1-371]), burst suppression on EEG on day 4 (8.2 [1.3-59]), and MRI deep gray matter injury (OR 33 [2.4-460]). These factors stratify neonates into low (0-1 factors; 3% [0%-14%] risk), medium (2 factors; 56% [21%-86%] risk), and high-risk groups (3 factors; 100% [29%-100%] risk) for epilepsy. The stratification was robust to varying clinical interpretations of the MRI and EEG. Neonates with hypoxic-ischemic encephalopathy who undergo selective head cooling appear at risk of epilepsy if they have 2 to 3 identified factors. If validated, this rule may help counsel families and identify children for close clinical follow-up.