Purpose: Is the capacity of visual short term memory limited by the number of objects that can be stored (4, Luck & Vogel,1997) or by the total amount of information? To address this question we estimated the memory capacity of a variety of materials using change detection and the information load of these same materials using visual search. Method: 6 subjects were tested on five stimulus classes (colors, polygons, Chinese characters, shaded cubes, and letters). In the change detection task, 1-15 items from one class were presented for 500 ms, followed by a 900 ms blank interval, and then by the test array. On half of the trials one item changed identity. The number of items yielding 75% correct detection of a change was used as the capacity estimate for that stimulus class. In the visual search task, a target item was presented for 500 ms, followed by a 900 ms blank interval, and then by the presentation of 4, 8, or 12 items of the same class, including the target on half the trials. The slope relating reaction time to set size for target-present trials was taken as the estimate of processing rate, and therefore information load per item, for each class. Here we assume that the more information that must be analyzed per item, the slower the search rate. Results: Memory capacity varied significantly over the 5 stimulus classes, ranging from 3.5 to 7 items. Visual search rate also varied markedly but in the opposite direction. Processing rate strongly predicted (r = .996) the inverse of capacity in each class. Conclusion: Memory capacity was not a constant in terms of number of objects but decreased as the information load (as indexed by search rate) of the items increased. Consequently, the capacity of visual short term memory is not fixed in terms of the number of objects, it is fixed in terms of the total amount of information (the product of the information load per item and the number of items that can be stored, minus a baseline capacity). Supported by NEI EY09258, Harvard Graduate Prize Fellowship.