Experiments 3 to 12 show that certain kinds of visuospatial material will differentially disrupt a visual memory strategy, even though subjects are instructed that the material is irrelevant. This irrelevant pictures effect is shown with a wide range of visual material including dynamic visual noise (Experiments 3 to 7 and 9 to 12), line drawings of common objects (Experiments 3 and 8), three by three matrices with the cells randomly black or white (Experiment 7) and a single dot that consecutively appears in five different positions in space (Experiment 12). The irrelevant pictures effect can also be demonstrated with two visual memory strategies that are qualitatively different from each other, the pegword mnemonic strategy and the method of loci strategy (Experiment 11). This robust irrelevant pictures effect is used as a tool for investigating visual working memory and the results of Experiments 1 to 12 are presented as being broadly consistent with Logie's 1995 two-component model of visual working memory comprising a visual cache and an inner scribe. The irrelevant pictures effect is evidence for the existence of the visual cache into which visual material is thought to have obligatory access. Experiment 9 demonstrates that static noise causes no interference with visual memory. Serially presented static noise in Experiment 10 does, however, selectively disrupt the visual strategy. It is hypothesised that the difference between the effect of the static pattern and the re-presented static pattern reflects the decay function of the visual cache. There is also support for the inner scribe part of the model. The star dots in Experiment 11 are thought to disrupt the pegword mnemonic strategy because both draw on common resources within the inner scribe. Moreover, Experiment 11 shows that the two components of Logie's 1995 model of visuospatial working memory can be empirically distinguished within one experiment. It shows that a visual task which requires access to the rehearsal component is disrupted by a concurrent, irrelevant spatial task as well as a concurrent, irrelevant visual task. In contrast, a visual task that requires little access to the rehearsal mechanism is not disrupted by a concurrent spatial task but is disrupted by a concurrent visual task. There are, however, also circumstances when the irrelevant visuospatial material disrupts visual and verbal memory strategies in a more general way, making attentional demands on the central executive part of working memory that are additional to those required to carry out the memory strategies. Three experiments look at this general effect of irrelevant pictures. Results suggest that the factors responsible for the placing of general demands on the central executive's attentional resources involve a focused, sufficiently unexpected change (Experiment 7) at the encoding stage of information processing (Experiment 8).