This paper examines the effectiveness electroencephalogram (EEG) as a biometric identification of individual subjects in a pool of 40 normal subjects. The EEG's second order statistics are computed using autoregressive models of various order. The coefficients in these models are then evaluated for their biometric potential. Discriminant functions applied to the model coefficients are used to examine the degree to which the subjects in the data pool can be identified. The results indicate that the EEG has significant biometric potential. In this data pool, 100% of subjects are correctly classified when all data is used, and over 80% when the functions are computed from half the data and then applied to the remaining.