This talk focuses on systems where the user enters text by pressing keys or buttons, and where more than one character or letter is assigned to each button. Such keyboards are ambiguous because there is uncertainty as to the intended symbol when a key is pressed. There is recent worldwide interest in such ambiguous keyboards because of mobile computing where space is limited. Also, such keyboards widen the communications possibility for users with physical disabilities who have insufficient motor facility to operate a full-size keyboard. We will first investigate the design possibilities. Many are variants of the phone keypad, while others use fewer keys (more ambiguity) or more keys (less ambiguity). Metrics to characterize such keyboards are defined and given, such as T-factor (the number of keys bearing letters) and KSPC (the average number of keystrokes to produce each character of text in a given language using a given interaction technique). Human performance issues are also noted, since disambiguation methods require additional attention for users. The promise of combining predictive techniques, such as word completion, with ambiguous keyboards is noted and demonstrated using software tools.