Visual comparison between different-sized objects with respect to shape can be done by encoding one of the objects as a mental image, transforming the image to the size format of the other object, and then testing for a match (Bundesen, C., & Larsen, A. . Visual transformation of size. Journal of Experimental Psychology: Human Perception and Performance, 1, 214-220). To identify the brain structures implicated in mental transformation of size, we measured the distribution of regional cerebral blood flow (rCBF) by positron emission tomography (PET) in 12 normal subjects who compared random stimulus patterns with respect to shape regardless of variations in size in a one-back match-to-sample paradigm. Each subject was PET-scanned 12 times during repetitive injections of H215O. In one condition (three scans), all stimulus patterns were small. In a second condition (three scans), all stimuli were large. In the third condition (six scans), the stimuli alternated between small and large. Mental transformation of size should occur in the alternating-size condition but not in the fixed-size conditions. As expected, behavioral measures (reaction time [RT], d', β) were nearly the same for the two fixed-size conditions but mean RT was longer and d' smaller in the alternating-size condition. Changes in rCBF specific to mental transformation of size were estimated by contrasting the alternating-size with the fixed-size conditions by use of statistical parametric mapping (SPM96) at a threshold of p < .05 corrected for multiple comparisons. The detected brain structures implicated in mental transformation of size were primarily located in the dorsal pathways, comprising structures in the occipital, parietal, and temporal transition zone (predominantly in the left hemisphere), posterior parietal cortex (bilaterally), area MT/V5 (left), and vermis (bilaterally). Contrasts between the two fixed-size conditions showed significant effects in only the occipital cortex.