This tidy addresses the question of whether there is value to the use of animation in icon design over and above that of r&acting aaention Specifically, the stody asks if there is any basis in human performawe for recommending for or against the we of animation in icon design. The objective was to examine the effect of animation on icon comprehension and preference at two levels of complexity, as measured in the context of a supermarket-shopping computer program. Results suggest that animation can contribute to users' comprehension of icons that represent commaads, and that this contribution may be greater for commands of higher levels of abstraction. Specifically, it appears that simple icons are more comprehensible than complex ones for commands of lower levels of abstraction, while complex icons are more comprehensible than simple ones for commands of higher levels of abstraction. Moreover, animation of complex icons that represent abstract commands may improve icon comprehension far more than animation of either simple or complex icons that represent less abstract commaads. The notion that animation can be used intelligently to improve human performance ia HCI is supported. Further research is encouraged to identify ways in which levels of command abstraction can be measured reliably, aad to establish guidelines for the level of abstraction at which complexity and animation should be introduced into icon design.