This chapter investigates the effectiveness of multimedia computer assisted instruction. Sixty-three first year university students were randomly assigned to 3 experimental groups receiving the same sequence of lessons, but with different presentation media: (i) 'Text'; (ii) Text and 'Stills'; and (iii) Text, Stills and 'Audio'. Wickens' (1992) model, an elaboration of dual code theory (Paivio, 1971), was used to investigate the 'why' of multimedia effectiveness. Knowledge type (declarative/procedural) and cognitive learning style (Verbal/Visual) were used to explore when and to whom multimedia is beneficial. Learning was assessed through a 20-item multiple-choice test, immediately following instruction, and 1 week later. Half the recall questions featured a graphic cue, so that the role of visual images in long-term recall could be examined. The mono-medium group (ie 'Text') demonstrated superior gains in learning over the other groups, with a trend for the 'Text' participants to have the highest recall test scores emerged, and was maintained across all hypotheses. The results were not in accordance with dual code theory, and challenged the widely accepted view that multimedia, especially graphic media, leads to increased learning.