The use of Somatosensory Evoked Potentials (SSEPs) has been an established electrophysiological tool for diagnosis of neurological disorders or injury. We use SSEP for the prognostication of outcomes after hypoxic-ischemic brain injury. Previous studies on rats with median nerve stimulation have primarily focused on short-latency SSEP within 30msec after stimulus. This study shows that long-latency SSEP (LL-SSEP) within 30-100msec is also of unique importance in monitoring brain injury induced by cardiac arrest (CA) and prediction of long-term recovery. In this study, 16 rats underwent either a 7min or 9min hypoxic CA. The Neurological Deficit Score (NDS) measured at 72hr post-CA was used to specify good outcome (NDS≥50) and poor outcome (NDS<50). Firstly, the LL-SSEP showed sharp responses to CA insults-change in P60 peak in the time-frequency space, Shannon entropy in the time domain, and wavelet entropy in the frequency domain. Secondly, LL-SSEP during early recovery had significant prognostic value: the Shannon entropy within 60min post-CA was higher for the good-outcome group (p-value=0.02, Student's t-test) and a delayed P60 exclusively predicted poor outcome. Thirdly, the LL-SSEP was significantly different than the short latency response. Since the LL-SSEP occurs well beyond the time delays for production by the thalamocortical network, it may be an independent cortical response, and may reflect the recovery of cortical neurons. The discovery of LL-SSEP should have significant clinical potential in assessing the recovery of the cortical function after brain injury and should be helpful in understanding the mechanism of thalamocortical arousal.