Likelihood inference is one of the classical methods for the estimation of model parameters, and it relies on the maximization of likelihood function of an assumed model based on the observed data. This chapter details the procedure of finding the maximum likelihood estimates of model parameters, mean lifetime, and reliability under normal operating conditions on the basis of one-shot device testing data under constant stress accelerated life-tests for different lifetime distributions. The EM algorithm involves two steps in each iteration of the numerical method of maximizing the likelihood function: expectation-step in which the missing data are approximated by their expected values, and maximization-step in which the likelihood function, with imputed values replacing the missing data, gets maximized. The chapter also discusses several associated interval estimation methods for all lifetime parameters of interest. Then, the results of a simulation study evaluating the performance of the developed point and interval estimation methods are presented.