Visual communication is an increasingly common part of environmental decision-making, being used as a ‘common currency’ to facilitate dialogue between policymakers and non-experts, increase understanding, and thereby improve the decisions made. GIS-based landscape visualisation is one method of producing images for consultation exercises, and while continuing advances in technology allow expansion into new areas such as the visualisation of rural landscapes, there is considerable feeling that we should not allow ourselves to be guided simply by what the technology can do. Instead, we should carefully assess whether each of these increases in capability can enhance the usefulness of visualisations. The need for a critical eye is particularly apparent when considering the advances in realism, since opportunities for realistic visualisations are rarely matched by the availability of suitably detailed data, and viewers’ perceptions of factors such as accuracy and certainty are also affected by the level of realism in an image. Following suggestions in the literature, the research described here attempts to determine whether there might be a ‘sufficient’ level of realism somewhat below the maximum possible level, and also begins to investigate the relative importance of various elements within an image such as buildings and foreground vegetation. Several visualisations of English rural scenes, with elements at varying levels of detail, were rated by survey respondents according to how well they felt they communicated the appearance of the landscape. The results do not show evidence of a ‘sufficient’ level of realism, but strongly indicate that some elements are more important than others. Foreground vegetation and the appearance of the ground surface over the whole scene were found to have significant effects on ratings, although the precise effect varied between the three scenes used, and much variation in ratings was not accounted for.