In this Paper we developed two discordancy tests D1 and D2 for detection of one and two upper outliers respectively for sized biased Exponential Distribution. To compare the performance of two newly developed methods we used two outlier detection methods based on moments sample skewness and kurtosis namely G1 and G2 respectively. The performance of four methods was compared by using simulation study. The results indicate that in case of single outlier the performance of method D1 is better as compare to all other methods. While in case of two outliers D2 method outperforms all other methods. At the end, we also determined the critical values of above four methods at 5% and 1% level of significance for n=5(1)20(5)40(10)100 by using simulation study