IntroductionThe goal of this study was to investigate the extent to which the informational variables from the preference matrix (Kaplan and Kaplan, 1989) are predictive of landscape aesthetics. In a recent meta-analysis covering empirical research on the preference matrix, it was concluded that consistent support was lacking for each of its component variables (Stamps, 2004). Possible causes for the inconclusive findings in the past are outlined and addressed in the present study. Amongst the methodological improvements are the use of a statistical test which takes into account the ordinal distribution of the data, improved definitions of some of the preference matrix items based on participant comprehension scores, a substantially sized image database (N = 1600) with high quality images from natural, built and “mixed” scenes, and use of beauty rather than preference as target variable since the latter is deemed to be sensitive to the goals and intentions of people (Herzog & Leverich, 2003). In addition, any confounding effects of scene familiarity and both natural and built content on ratings of the variables from the preference matrix were taken into account and the frequently suggested interaction between complexity and coherence was investigated.Method Results and Discussion In contrast to findings reported previously (Stamps, 2004), this research shows that all variables from the preference are positively predictive of beauty. The lack of consistent findings in previous studies is therefore likely a consequence of methodological shortcomings. In addition, this study has been the first in which support for an interaction between coherence and complexity was found. Furthermore, both natural and built content of a scene, as well as its familiarity, were found to be related to reported beauty. This finding is interesting because it implies that natural and built content are concepts that are not directly opposite. It is concluded that the preference matrix should be regarded a valuable paradigm for informing future research that has a focus on human preference for specific kinds of spatial information.