An improved email classification method based on Artificial Immune System is proposed in this paper to develop an immune based system by using the immune learning, immune memory in solving complex problems in spam detection. An optimized technique for e-mail classification is accomplished by distinguishing the characteristics of spam and non-spam that is been acquired from trained data set. These extracted features of spam and non-spam are then combined to make a single detector, therefore reducing the false rate. (Non-spam that were wrongly classified as spam). Effectiveness of our technique in decreasing the false rate shall be demonstrated by the result that will be acquired.