Subjects completed a menu selection task using mouse and spoken language technologies for menu sizes that are powers of 2 from 2 to 256. It was found that mouse selection was faster than spoken language selection for menu sizes of 32 or less for most subjects. For small N, mouse selection was about three times faster. Spoken language selection was as fast or faster for menus of size 64 or larger. The trade-off size, approximately 64, was different for various subjects, but not greatly different. Data was gathered for words with one, two, and three syllables with the result that the trade-off size was the same for all three, but the error characteristics for spoken inputs were worse for shorter words, as would be expected.