The major issues in text entry on mobile computer systems are presented. Two formal usability experiments were conducted to explore particular problems in the design and usability of high performance soft keyboards. An experiment with 12 participants compared text entry rates on large and small soft keyboards with QWERTY layouts. Participants were also given a standard touch typing test. The results showed that touch typing skills on physical keyboards do not necessarily transfer to stylus tapping on a soft keyboard. Participants touch typing speeds ranged from 19 to 74 wpm, whereas their stylus tapping speeds ranged from 15.4 to 25.0 wpm on a large soft keyboard and from 13.1 to 26.0 wpm on a small soft keyboard. The correlations were low:.416 for the large soft keyboard and.523 for the small soft keyboard. We also designed a completely new high performance soft keyboard, based on a linguistic model, Fitts' law, and other models. The new design supports a predicted maximum text entry speed of 58.2 wpm. This is about 35% faster than with a QWERTY layout (at a maximum of about 43.2 wpm). A second longitudinal text-entry experiment with five participants was conducted to compare our high performance soft keyboard with a QWERTY soft keyboard. (Abstract shortened by UMI.)