Prolonged EEG monitoring in the ICU is now common. Studies suggest that the prevalence of seizures and nonconvulsive status epilepticus (NCSE) is 10% or more. However, the selection criteria for monitoring remain ambiguous, and the impact of prolonged EEG monitoring on treatment is unclear. We addressed these questions by performing prolonged EEG on patients for whom a routine 30-min or prolonged EEG was requested by ICU staff. During a prospective but not randomized 22-month study, EEGs were requested by ICU staff on 130 patients, with diagnoses of trauma, tumor, stroke, metabolic derangement, cerebral hemorrhage, and hypoxic–ischemic injury. Patients were placed in three groups. Group 1 : 34 patients received a 30-min EEG, because for various reasons long-term recording could not be done. Group 2 : 83 patients for whom staff requested a 30-min EEG, instead received 16–24 h of continuous video-EEG monitoring. Group 3 : 13 patients, with known epilepsy, for whom continuous video-EEG monitoring was requested, received 16–24 h of recording to monitor suspected status epilepticus. Epileptologists compared the first 30 min of EEG with the subsequent recording to see if any additional information was obtained, and if it impacted treatment. Group 1: 34 patients had a routine 30-min EEG. One was normal, 27 were slow or poorly reactive, and six (18%) showed epileptiform activity, including one with electrographic seizures. Group 2: 83 patients were monitored with video-EEG for 16–24 h. All EEGs were abnormal with slowing and poor reactivity. 28/83 patients (34%) showed epileptiform findings in the first 30 min, including periodic epileptiform discharges (PEDs), generalized or focal epileptiform discharges, burst suppression, triphasic waves, and 2 patients had clinical seizures. 5/28 developed additional epileptiform changes overnight, including two more with clinical seizures. 55/83 patients (66%) had no epileptiform findings in the first 30 min, but 7/55 developed these overnight, including two with electrographic and one with clinical seizures. Overall, in 12/83 patients additional epileptiform abnormalities developed overnight, and in 7/83 treatment was changed based on prolonged as opposed to routine EEG. 3/7 patients in whom treatment was changed showed improvement. Group 3: 13 patients known to have epilepsy, who presented with seizures, were deliberately placed on long term video-EEG monitor. 9/13 (69%) showed epileptiform abnormalities in the first 30 min including three with NCSE and one with focal seizures. Overnight two more evolved into NCSE, and prolonged EEG influenced therapy in 6/13. In 83 unselected ICU patients, overnight video-EEG, as opposed to a 30-min EEG, detected additional epileptiform abnormalities in 12 patients (14%), and 7/83 patients (8%) had changes in treatment based on findings from prolonged as opposed to routine EEG. In contrast, patients with epilepsy who presented with seizures were more likely than other ICU patients to have NCSE, and prolonged EEG influenced treatment in 6/13 (46%). This study suggests that long-term EEG monitoring in an unselected ICU population rarely influences treatment or outcome. In a selected population with known epilepsy the benefit may be larger.