Consider a series system consisting of k-independent components. Assuming t
hat the life time of the ith component follows an exponential distribution
with failure rate Xi, i = 1, 2,..., k, we address the problem of point esti
mation of the reliability of the system at a certain time point t > 0. Moti
vated by a numerical study, we investigate the second order risk of the max
imum likelihood estimator of the system reliability and establish its secon
d order admissibility. We have also studied the unique minimum variance unb
iased estimator and two popular generalized Bayes' estimators which turn ou
t to be inferior to the MLE in terms of the first order mean squared error.