OBJECTIVE To measure association between paroxysmal events and length of monitoring to identify a practicable duration time for capturing seizures in the elderly. METHODS Consecutive inpatients 60 years and older who were admitted to the Epilepsy Center and underwent prolonged video electroencephalogram (VEEG) monitoring (VEM) were reviewed retrospectively. Electronic medical records were reviewed to collect information regarding sex, age at onset of symptoms and examination, concurrent epilepsy, frequency of seizures, diagnosis before and after examination, antiepileptic drugs (AEDs), brain magnetic resonance imaging (MRI), and VEEG findings. RESULTS A total of 184 consecutive elderly inpatients were enrolled. The mean age was 67.1±6.1 years (range, 60-89 years), with 69 females and 115 males. Mean length of monitoring was 20.4±18.9h (range, 1h-6days). During LTM, 89 patients (48.4%) recorded paroxysmal events, including 58 epileptic seizures (43.3%) and 31 non-epileptic events (16.8%). All non-epileptic events were captured during the first 24h. All first epileptic events were detected during the first 4days, with 98.9% of them recorded by the end of the 2nd day. Increased seizure incidence (p=0.000, odd ratio [OR]=0.075, 95% confidence interval [95%CI]: 0.035-0.163) and length of monitoring (p=0.001, OR=1.044, 95%CI: 1.017-1.071) were independently associated with paroxysmal events capture. CONCLUSIONS It may be practicable to monitor for 24h when a non-epileptic seizure is suspected, with expected monitoring duration of 2days when an epileptic seizure with daily or persistent frequency is considered, except for pre-surgical evaluations.