When young children draw a scene in which one object partially occludes another, they often depict hidden features of the occluded object by rendering the nearer object transparent. This may signify a motive to produce drawings that are informative. Three experiments are reported in which a simple paradigm is employed to consider this possible basis of transparency drawing. Experiment 1 established a gradual decrease in the use of this device across the 5–7‐year age range. Its occurrence could not be ascribed to a simple lack of graphic skill and it was not readily inhibited by stronger perceptual marking of scene boundaries. The extent to which children were purposefully producing informative drawings was evaluated in Expts 2 and 3. Because an increase in occluded information actually served to inhibit transparencies, it is argued that they need not reflect such a communicative attitude.An alternative account is proposed whereby drawing decisions are guided, sometimes inappropriately, by the structure of cognitive representations children form of scenes. Transparencies signify a failure to anticipate certain graphic ambiguities thereby generated. However, they may be inhibited if the geometry of a particular scene forces the child to confront such ambiguity at the outset of drawing.