Beginning with a survey of work previously done on the subject, this study attempts to learn more about how, and to what extent, children benefit from pictorial illustrations, with a view to improving instructional materials. Three areas were investigated with experiments using children from nursery schools and kindergarten to sixth grade. The first experiment probed the effect redundant clues have on the recognizability of an illustration, the second the relative values of realistic and abstract illustrations in the teaching of a concept, and the third the effect of age differences on the perception of pictures. The first experiment, in which the pictures were presented by means of a tachistoscope, led to the conclusion that a picture becomes more recognizable as more clues are furnished. In the second experiment, children taught the concept "one-half" by means of abstract materials did as well as those taught with illustrations of real objects: a four-year-old's idea of what is abstract is, apparently, not an adult's. The best learning was achieved when children were taught and tested with realistic materials. Children learned best when the illustrations suggested a kinetic situation. Perception increased with age. A great deal more research, though apt to be expensive, is recommended. (GO)