Abstract: Electroencephalographs are records of brain electrical activity. It is an indispensable tool for diagnosing epileptic seizure. Manually reviewing EEG recordings for detection of epilepsy pattern is a time-consuming process. It is, therefore, necessary to automate the epileptic seizure detection. Further analyzing after the detection using large data set is a good supplement to the wide range of algorithms currently used for analysis. Seizure evolution is typically a dynamic and non stationary process and the EEG signals are composed of many frequency bands. The objective of this work was to determine features that differentiated epileptic seizure from a normal activity. Subjects suffering from a commonly occurring generalized epileptic seizure EEG segments and non seizure EEG segments are used for the study. The Spectral and Statistical methods are applied to the signal and the features are extracted. Spectral power, Standard deviation, Variance, Root Mean Square and Measure of Spread are the features that differentiate abnormal activity from normal activity whereas Median, Mode, Skewness, kurtosis are not able to differentiate. This study gives a method to detection of seizure and analysis in offline. It can be further extended to the real time. The algorithm is tested with two different databases covering children and as well as adult data sets.