Subjects rated the perceived similarity between patterns and their 180-deg rotational variants. The stimuli, adapted from Halpern, Fishbein, and Warm (1979), were randomly generated dot patterns and line drawings and polygons generated from the dot patterns. Inspection time was varied in a between-subject design. As in the Halpern et al. (1979) study, it was found that the variables of form type (dots,lines, polygons) and axis of rotation (x, y, z) had no effect on perceived similarity in the long exposure conditions. However, when inspection time was limited to 150 msec, both variables had a substantial impact. This finding undermines the frequent practice of selecting stimuli for use in reaction time or other experiments with limited inspection time on the basis of similarity ratings gathered under typical free inspection conditions. How similar do two stimuli appear to an observer? What influence does a selected variable have on the pattern of perceived similarity within a set of stimuli? These questions are routinely posed by psychologists. In studies of visual perception, the answer is usually obtained by allowing an observer to freely examine a stimulus pair until he is ready to make his rating response. Information from a series of such judgments may be used to test a model of perception when that model makes predictions about perceived similarity. Similarity ratings may also be used in the stimulus selection phase of an experiment in order to equate similarity across stimulus sets, or vary it methodically. The purpose of this short report is to bring to the attention of psychologists who use similarity ratings our finding that stimulus duration has a profound effect on the pattern of perceived similarity. More specifically, at very short exposures (150 msec), which are rarely used in ratings studies, perceived similarity is very sensitive to variables that have no effect at longer exposures. This finding undermines the common practice of selecting stimuli for use in reaction time and other experiments with brief exposure durations on the basis of similarity ratings gathered under free inspection conditions.The stimulus patterns we used were taken from a study by Halpern, Fishbein, and Warm (1979) and are shown in Figure 1. These investigators were interested in the problem of representativeness in the perceived similarity of patterns. Their experiment compared the rated similarity of transformed stimuli Thanks are extended to D. Halpern for providing the stimulus patterns used in this study.