Variation has in recent years become an important concept in the analysis of loads on the locomotor system in occupational life. Studies of controlled variation in isometric exercise have concentrated on the effects of introducing pauses at regular intervals into an otherwise constant load. A complete description of the cyclic character of such regimens requires a set of three parameters. We suggest the use of the set mean load/cycle time/duty cycle. The design of laboratory studies should more often be based upon the activity patterns found in occupational life. In studies observing exposure, the variation concept introduces problems concerning how to define and how to quantify 'variation'. It is shown that present methods (APDF analysis and contraction frequency analysis) individually fail to pick up important aspects of variation. A new method for exposure variation analysis (EVA) is proposed. Examples are presented of EVA performed on vocational EMG registrations.