Abstract: Essential genes are indispensable for the maintenance of life activity. Gene feature selecting is crucial for the theoretical predictions and recognition of essential genes. In this paper, based on chaos game representation method, comparative analysis of the Hurst exponent of the essential and non-essential genes of all bacterials in DEG essential genes database was made. The result shows that both essential and non-essential gene sequences have long range correlation. There is a linear relationship between Hurst exponent and gene sequences, and the Hurst exponents of essential and non-essential genes have significant difference. Therefore, Hurst exponent may be an effective feature for distinguishing essential and non-essential genes. The results provide theoretical basis for bacterial species essential genes prediction.