A crucial aspect of shape similarity estimation is the identification of perceptually significant image elements. In order to understand more about the process of human segmentation of abstract images, a sample of 63 trademark images was shown to several groups of students in two experiments. Students were first presented with printed versions of a number of abstract trademark images, and invited to sketch their preferred segmentation of each image. A second group of students was then shown each image, plus its set of alternative segmentations, and invited to rank each alternative in order of preference. Our results suggest that most participants used a relatively small number of segmentation strategies, reflecting well-known psychological principles. Agreement between human image segmentations and those generated by the ARTISAN trademark retrieval system was quite limited; the most common causes of discrepancy were failure to handle texture and incorrect grouping of components into regions. Ways of improving ARTISAN'S ability to model human segmentation behavior are discussed.