We investigated the effect of various strategy-modeling techniques on the performance of both young children and elderly adults on the 20 Questions Task. In experiment I, 6-year-old children and elderly adults received training in either how to classify the stimuli, how to ask constraint-seeking questions, or how to use the information gained from the answers to their questions. Only training in how to ask constraint-seeking questions significantly facilitated performance. In experiment II, 4- to 6-year-old children and elderly adults were given training that was successful in experiment I: providing an example of a constraint-seeking question and providing the rule for generating a constraint-seeking question. Both types of training facilitated the performance of both young children and elderly adults.