Electromyographic activity of the upper trapezius muscles was recorded over the workday for two groups of service workers, shopping centre (n = 22) and healthcare workers (n = 44), both with low observed biomechanical exposure. Static and median EMG activity level, number of EMG gaps and gap time were determined. The variability of these variables over the workday was examined by calculating the coefficient of variation (CV) and the intraclass correlation coefficient (ICC) of 1-h consecutive recording periods. All variables except gap time showed acceptable reliability (ICC = 0.69-0.78), i.e. the largest fraction of variance in the data set was due to intersubject variance, despite relatively large hour-to-hour variation (CV = 0.21 0.62). The EMG activity level in the trapezius muscles was low (static activity level < 1% EMGmax), despite the high prevalence of shoulder and neck pain for both groups of workers. In addition to the work recordings, tests were performed to determine intersubject variation in muscle activity when adopting a standardized resting posture, and in a dynamic muscle activity pattern during paced arm movement. Neither the EMG variables from the work recordings nor the tests with EMG recording indicated higher trapezius EMG activity levels for workers with pain in the shoulders and neck in this study. The low EMG levels are interpreted to indicate a low risk of developing shoulder and neck complaints due to biomechanical exposure for both groups of workers. The possibility of pain-initiating mechanisms, associated with stress and not mediated through muscle activity, is considered in the discussion.