This thesis uses visual imagery tasks (mental rotation and mental subtraction) to examine verbal interference and verbal facilitation in visual memory. It demonstrates how task demands can mediate the verbal interference and verbal facilitation effects in visual imagery. Using the mental rotation paradigm, this thesis places a focus upon the method of stimulus presentation during the learning phase and the test. It demonstrates how a presentation method that emphasizes serial order (the temporal presentation method) can elicit positive effects of covert spontaneous naming during both encoding and retrieval. In contrast, a presentation method that emphasizes spatial information does not show a significant role for covert spontaneous naming during encoding or retrieval. Further, under temporal presentation conditions, explicit labelling during encoding (via the use of either self-generated or experimenter-generated labels) is found to show an interfering effect compared to covert spontaneous naming. Using experimenter-generated labels, it is found that re-presenting the explicit verbal labels as cues at retrieval removes the interfering effects of explicit labelling during encoding and enhances performance. In addition, reducing exposure to explicit verbal labels during encoding is found to be a possible method for removing the negative effect of explicit verbal labels during encoding. Finally, the positive effect of covert spontaneous naming and the negative effect of explicit labelling are replicated using a different mental subtraction paradigm. Overall, the findings indicate that task demands determine the role of the verbal code in visual imagery. Hence, there is no unified theory to account for the role of the verbal code in visual memory, but different theories can be applied under different conditions.