The paper considers a selection goal for selecting the exponential population with the largest location from the set of ‘good’ popoulations which have location among the a largest ones and scale among the b largest ones. If that set is empty the goal is to select no population. In order to achieve the goal exactly, the paper gives different two-stage selection procedures for the two cases: (I) the set of ‘good’ populations cannot be empty, and (II) otherwise. The probabilities of a correct selection associated with these two-stage procedures are deived and the constants required to implement the procedures are computed by Monte Carlo method. A convenient method for determining an optimal initial sample size in the two-stage procedures is given by using tables of the constants for some initial sample sizes and for specified parameters. The efficiency of those procedures is discussed in relation to the specified parameters.