Three different strategies of choosing items for presentation in a simple list learning situation are compared. The instructional task was to teach the correct response to a number of stimulus items, using a paired-associate teaching procedure. Only one item could be presented on a given trial and the total number of trials was limited. The optimization problem considered was to find the best strategy for deciding which item to present a subject on a given trial, based on his performance on previous trials. The optimization criterion was the number of items retained on a posttest. Three progressively more sophisticated models of the learning process, the linear model, the one-element model, and a forgetting model, suggested different candidates for the optimal procedure. This paper is devoted to the derivation of theoretical expressions for operating characteristics of interest for each of the three strategies. These expressions, based on the most adequate model (the forgetting model), then permit direct numerical comparisons of the predicted performance of the three strategies for specified values of the model parameter. If the material to be learned was relatively easy, then the theoretical differences between the strategies were relatively insignificant. If the material to be learned was difficult, then the theoretical differences between the strategies were quite significant. The differences depended on the parameter values of the forgetting model in a systematic way. (Author/DT)