This study was conducted in order to begin to understand the effects of providing different types of visuals -- such as pictures, diagrams, and still photos of speakers -- with items on the computer based Test of English as a Foreign Language (TOEFL). A nested crossover design (subjects nest in proficiency, level, and form) was used to examine the effects of language proficiency (high or low), still photos (present or absent), and type of stimuli (dialogues/short conversations, academic discussions, mini-talks with context visuals, mini-talks with content visuals) on performance on TOEFL multiple-choice items. Three two-way interactions were significant: proficiency by type of stimuli, type of stimuli by visual condition and type of stimuli by time. The weakest of these interactions, type of stimuli by visual condition, was the most interesting; it indicated that there presence of visuals results in facilitation of performance when the visuals bear information that complement the audio portion of the stimulus. The interaction effect between type of stimulus and time suggests that the trend toward the use of longer stimuli is not without consequences, and these consequences may extend to testing situations. Responses to a series of questionnaires indicate that the majority of the subjects preferred the stimuli that were accompanied by visuals.