INTERACTION DESIGN REACHES DEEP into a prod-uct. It explicates system constraints, supports user goals and tasks, facilitates cognition, enables perception, and requires human action. The resulting user experience also extends beyond skin deep: It directly involves human musculoskeletal and nervous systems. As such, it can do them damage.We rely on "official standards," such as the definition of usability from ISO 9241, Part 11: "The extent to which a product can be used by specified users to achieve specified goals with effectiveness, efficiency and satisfaction in a specified context of use."But how easily can we test such usability? We need the highest return on our testing resources, so we mostly test with neophytes in sessions of an hour or two. We make sure our designs match users' goals, tasks, and mental models.We verify that our interfaces are explorable, effective, efficient, and even pleasing to our test participants.Yet our specified users and contexts of use include people exercising our designs over long periods of time.Will our testing find the consequences of long-term use?How much satisfaction has our design provided a "specified user" if using it causes a musculoskeletal disorder (MSD)? In addition to providing effectiveness, does our design efficiently strain connective tissue and crush nerves? We know about MSDs. We all know someone who has carpal-tunnel syndrome or other damage that involves chronic pain and impairment. The four factors of MSD-repetition, force, posture, and vibration-constitute the potentially hazardous attributes of every physical interaction we design, in software or hardware (with the possible exception of vibration... while haptic feedback remains scarce). If interactions damage users, the designs of those interactions bear partial responsibility. Few interaction designs have proved as deadly as the Therac-25 radiation-therapy machine interface, which killed three patients in the mid-1980s, or Multidata radiation machines, which apparently killed five more in the 21st century. However, less dramatic user experiences deliver far more victims. In 2003, according to the U.S. Bureau of Labor Statistics, MSDs of all types and causes accounted for 435,180 injuries and illnesses with days away from work-33 percent of all such injuries and illnesses [1]. But what about MSDs and human-computer interfaces? Countless studies have targeted that question. One such study [2] observed, "More than 50 percent of computer users reported musculoskeletal symptoms during the first year after starting a new job." As devices and interactions proliferate, our creation of MSDs constitutes neither a small nor a shrinking problem. Neither can we dismiss it as a problem of only lowprofile organizations. Take the Apple mouse, for example. This device shipped with my beloved Macintosh G5. : / 60 i n t e r a c t i o n s / j u l y + a u g u s t 2 0 0 5