Three experiments using the same overall design were conducted to address problems associated with repeated measurement designs employed to assess retention of information in complex pictures and to assess the developmental course of schemata-guided retention efforts. Forty-eight subjects, ages 6, 10, and 20 years, were shown scenes whose forms were varied between groups--arranged naturally, in quadrants, or vertically. Scene content for the four items in each scene, either high or low probability, was varied within subjects. In experiment 1, subjects were asked to recognize target components on an immediate recognition test containing equal numbers of high and low probability targets and distractors. In experiment 2, subjects were asked to recall the target item when the three items accompanying it during acquisition were represented in their original structural form during cued recall. Experiment 3 examined free recall. Results of all three studies showed developmental improvements in recognition. The effects of form and content did not vary developmentally when memory was tested by recognition or free recall. Naturalistic forms facilitated recall but not recognition at all ages. Age and content interacted when memory was tested by cued recall. (Test items are included.) (JL)