Abstract: In current energy detection algorithms, the environmental noise level is often assumed as prior information. However, it is a parameter which needs to be measured in practical application. In the situation of using the known free spectrum to measure noise power is impracticably, this paper proposes a novel broadband spectrum energy detection algorithm based on iteration which can make effective detection to the broadband sampling signal which prior information is total blind. This paper also demonstrates that when the initial value of the iterative algorithm is reasonable, the algorithm can get stable convergence. The simulation results show that the algorithm can obtain a very good performance if the length of time observation is more than 20.