Objectives: Our objectives were to address some key issues- to determine the minimal efficient duration of the standard and sleep EEG recordings, to assess the diagnostic yield of hyperventilation (HV) and to assess whether injuries occur more often in an Epilepsy Monitoring Unit (EMU) where portable EEG amplifiers are used, and where patients can freely move within a large area during the monitoring. Methods: For the first part of the study, we have reviewed 1005 EEG recordings and determined the shortest recording duration necessary to identify interictal EEG abnormalities. In order to assess the diagnostic yield of 5 min HV compared to 3 min HV, data were evaluated from 1084 consecutive patients, from three European centers, referred to EEG on suspicion of epilepsy. Seizures and interictal EEG abnormalities precipitated during the first 3 min and during the last 2 min of the HV period (totally 5 min) were determined. For the last part of our study, patients were monitored at the Danish Epilepsy Center. Adverse events (AEs) including injuries, were prospectively noted, as part of the safety policy of the hospital. Other data were retrospectively extracted from the electronic database, for a 5-year period (Jan 2012–Dec 2016). Results: Standard, awake recordings shorter than 20 min yielded a significantly lower incidence of abnormal findings as compared to longer recordings. Although there was an increase in the diagnostic yield from 30 to 180 min recording duration, this failed to reach the level of significance. For sleep recordings, there was no significant increase in the diagnostic yield beyond 30 min. 877 (81%) completed 5 min HV. Seizures were precipitated during the first 3 min of HV in 21 patients, and during the last 2 min in four more patients. Interictal EEG abnormalities were precipitated in the first 3 min of HV in 16 patients, and during the last 2 min in 7 more patients. Psychogenic nonepileptic seizures occurred in eight patients during the first 3 min of HV and in two more patients during the last 2 min. No adverse events occurred during the last 2 min of HV. 976 patients were admitted to the EMU. Falls occurred in 19 patients (1.9%) but none of them resulted in injury. The rate of AEs were similar or lower than previously reported by other centers, where the mobility of the patients had been restricted during monitoring. Conclusions: Our results provide evidence for recommending at least 20 min recording duration for standard awake EEGs and 30 min for sleep EEG recordings. 16% of seizures and 30% of interictal EEG abnormalities triggered by HV occurred during the last 2 min of HV, suggesting the clinical usefulness of prolonged hyperventilation for 5 min. The vast majority of patients (99%) who are able to hyperventilate for 3 min can complete 5 min HV, without additional adverse events. In an EMU specially designed for this purpose, where patients are under continuous surveillance by personnel dedicated to the EMU, injuries can be avoided even when the mobility of the patients is not restricted.