A new methodology for testing preschool children's comprehension of television is described and the results of the first experiment with this method are presented. Original program material was created by filming 30 second animated stories in color and transferring them to videotape for subsequent editing and addition of sound. Thirty-five 3-year-old and 36 5-year-old children viewed the stories. Their reconstructions of the depicted materials were taped and scored. Analysis of total scores revealed an effect of age such that 5-year-olds reproduced more story units than 3-year-olds. However, the reconstruction patterns were very similar for both ages. Visual units were more likely to be reconstructed than auditory units. There was no effect of editing condition--(1) no edits, (2) zooms and pans, (3) cuts, zooms and pans--on total scores, though hints of an editing effect showed in the reconstruction protocols, such that the unit edited out was less likely to be reproduced. A substantial proportion of 3-year-olds reproduced the gist of the story, while most 5-year-olds gave connected, accurate reconstructions. This methodology offers both control over program variables and information about the quality and integrated nature of preschool children's representations of television content. (Author/CM)