Which factors can predict aesthetic preferences for images of neatly organized compositions, collected on blogs like Things Organized Neatly© (http://thingsorganizedneatly.tumblr.com/)? In this project, we focused on both stimulus and person properties related to order, complexity, and the balance between order and complexity. Participants (N = 415) indicated their preference for one of two simultaneously presented images (100 pairs) and completed some personality questionnaires. In a second (optional) part of the study, participants (N = 84) rated how ordered, complex, soothing, and fascinating they perceived each of 184 images to be. Additionally, some objective statistical measures were calculated on the images (i.e., PHOG-derived measures of self-similarity, complexity, and anisotropy, Fourier slope, and fractal dimension). Amongst others, the proportion of participants that preferred a certain image in a pair related to differences in average fascination ratings between the images. The bigger the difference in average fascination scores was between the images in a pair, the larger the proportion of participants that preferred the most fascinating image in the pair. Subsequently, average fascination ratings for the images could be predicted by the average ratings for order and complexity (Adj–R² = .599). Interestingly, subjective ratings of perceived complexity correlated highly with some of the more objective statistical image measures, like fractal dimension (r = .63, p < .0001) and PHOG complexity (r = .62, p < .0001). Also, the moderate to high correlations between the different statistical image measures were similar to the correlations found in previous studies for different types of stimuli.