Recently there have been emerging interests and researches about Blind Source Separation problem. This problem is classical but known as difficult problem. Given the mixture of some source signal, we want to find out what the source signal is. The only assumption is the statistical independence of the source signal and we don't know what the source signal is and what the mixture is. Researchers in this field have developed many novel algorithms to access this problem. But the problem is that the separation capabilities of many algorithms depend on the properties of the source signal. That is, some algorithms separate well sub-gaussian signal but not super-gaussian signal and some algorithms vice-versa. In this paper, we deal with the BSS problem in the mixture of super-gaussian and sub-gaussian sources based on ME approach. The mixing is linear mixing and there is no time delay.