Some e-learning describes graphics using words in both on-screen text and audio narration in which the audio repeats the text. We call this technique redundant on-screen text because the printed text (on-screen text) is redundant with the spoken text (narration or audio). In this chapter, we summarize empirical evidence showing that people learn better from concurrent graphics and audio than from concurrent graphics, audio, and on-screen text. We update research and theory that has appeared since the previous edition of this book, but the overall message remains the same: In general, do not add printed text to a narrated graphic. The psychological advantage of presenting words in audio alone is that you avoid overloading the visual channel of working memory that can occur when the eyes focus both on graphics and on printed words. There are also certain situations that benefit from the use of redundant on-screen text, which we call boundary conditions. We describe those here as well, including adding printed text to narration when (1) there are no graphics, (2) the presentation rate is slow paced or learner paced, (3) the narration includes technical or unfamiliar words, and (4) the added text is shorter than the narration.