Recently, Kuchinke, Trapp, Jacobs, and Leder (2009) used pupillary peak dilations (PD) to test the hypothesis that fluent picture processing elicits aesthetic affects. They used reproductions of cubist pictures of different abstractness as stimuli, which was assumed to modulate processing fluency. As a result, less abstract pictures were not only processed more fluently and preferred, they also produced larger PDs than more abstract ones. This was interpreted as support of their hypothesis. The aim of the present study was to replicate and generalize these results with an improved method and by adding expressionist pictures, which covered a relatively large range of abstractness. In the first experiment, where art style was blocked, there were no clear results. Therefore, we randomized art style in the second experiment. This time PDs increased with decreasing abstractness, even though significantly only for the expressionist pictures. However, there was no relation between preference and PDs. Thus, although we also observed a covariation between abstractness and pupil size, our data do not support the idea that PDs reflect fluency-induced aesthetic affect.