Purpose: While previous research has shown that the orientation of a single grating can be stored in visual short-term memory with high fidelity (Magnussen, Idas, & Myhre, 1998), the following experiment demonstrates that the maximum number of orientations that can be remembered is lower for Gabor patches (sinusoidal gratings attenuated by a gaussian envelope) than for lines. Method: Memory capacity for orientation was estimated in separate blocks of trials for Gabor patches (high contrast gratings with a spatial frequency of 1 cpd and diameter 3 deg) and lines (a single black bar of a Gabor patch subtending .4 deg by 3 deg). On each trial 1-5 Gabors or lines of various orientations were briefly presented, followed by a blank interval, and then by a second presentation of objects. On half of the trials the two displays were identical, and on the other half one of the items changed orientation by 90 degrees. The task was to indicate whether or not any of the items changed orientation. Results: Averaged across 8 subjects, change detection accuracy was not significantly different for Gabors and lines at set size 1 (accuracy was greater than 95% for both). However, accuracy was significantly lower for Gabors compared to lines at larger set sizes, with an estimated capacity of approximately 2 for Gabor orientations but 3 for line orientations (t(7)=5.7, p<.001). Conclusion: Although a change in the orientation of a single Gabor can be detected as accurately as a change in the orientation of a single line, the maximum number of individual orientations that can be remembered is lower for Gabors than for lines. These findings suggest that orientation cannot be stored as an abstract code independently of the object that carries that orientation. Thus, visual short-term memory capacity for orientation is not fixed in terms of the number of orientations that can be stored, but depends on the type of objects to be remembered.