As a list of menu items becomes longer, users have increasing difficulty finding the desired item. Researchers have noticed that some menu items are selected more frequently than others, and have suggested various kinds of adaptive menus that support the selection of these high-frequency items. In this paper, we investigate existing selectionsupportive menus and find that they support the selection of high-frequency items by providing them with either spatial or spatio-temporal priority. For example, split menus (ACM Trans. Comput. Human Interaction 1 (1994) 27-51) offer spatial priority and the menus of Microsoft Office 2000 provides both spatial and temporal priority. This finding leads us to define a new type of adaptive menu that provides only temporal priority, where only high-frequency items are displayed at first. A controlled experiment was performed to assess the issues involved in the design of adaptive menus. The results revealed that each type of adaptive menu has both strengths and weaknesses. Spatial prioritization significantly decreases the selection time of high-frequency items if the variations in selection frequency are small; however, if the selection frequency distribution changes greatly, then the mean selection time rapidly deteriorates. In comparison, temporal selection- support is less effective at reducing selection time for high-frequency items, but is more robust due to its insensitivity to variations in selection frequency. Based on experimental data, quantitative criteria are provided to assist designers when deciding which type of menu to use.