1. Experienced touch typists were asked to type words that contained only one or two letters typed by one of the two hands. When a word contained a pair of letters typed by one hand, the letters could be consecutive, or there could be one, two, or three intervening letters typed with the other hand. 2. We studied cases in which pairs of letters were either identical, different but typed with the same finger, or typed with two different fingers on the same hand. 3. Translational and rotational motion of the fingers and wrist was computed optoelectronically from the location of reflective markers on the hands. Finger and wrist motion recorded when subjects typed pairs of letters was compared with the motion recorded when the subject typed either letter in isolation. 4. When the subject typed the same letter consecutively, or separated by intervening letters, the second keystroke began only after the first key had been pressed. The same result was obtained when the second letter was not identical but was typed with the same finger. Up to the time of the first keypress, the initial keystroke kinematics were identical to those for that letter typed in isolation. 5. When the second letter in a pair was typed with the use of a different finger, the initial focal movement (wrist and finger striking the key) was unaffected up to the time of initial keypress. However, the second finger could begin to move toward the second key shortly before the initial keypress, and therefore the corollary movements normally involved in the initial keystroke were affected. 6. These results indicate that typing movements are executed primarily in a serial fashion, letter by letter. There can be some overlap between consecutive keystrokes only if they are executed with different fingers. 7. Words in which two letters typed with one hand were separated by three letters typed with the other hand provided subjects the opportunity to initiate the second keystroke at a range of times after the first keypress. 8. When the second letter differed from the first, subjects always returned to the home position after the first keypress and initiated the second keystroke with a normal latency. However, when the second letter was the same as the first letter, subjects sometimes suppressed the return to the home position after the first keystroke and maintained their finger poised over the key. 9. Thus keystrokes of one hand are best described as being executed sequentially. However, the findings presented here also indicate that movement planning encompasses strings of letters.