While electromyography (EM G) is commonly used to quantify levels of muscle activity in different jobs and tasks, less attention has been paid to the size and sources of exposure variability between and within individuals at work. In particular , few attempts have been made to discriminate between biological sources of variabilityand methodological uncertainty, including the effects of the EMG normalization procedure. The present study aimed to quantify the testretest repeatability of reference voluntary effort (R VE) contractions of the upper trapezius and the amount of variance introduced to EMG exposure parameters due to normalization trial variance.

EMG data was collected from the right trapezius muscle of 15 female subjects during 36 work cycles of simulated industrial assembly work on t wo separate days. For all cycles, EMG was norrnalized to each of the four RVE calibration trials collected on the day. From the normalized data the following exposure parameters were calculated for all cycles and norrnalizations: 1 St, 10th, 50th, 90th, and 99th percentile APDF values; mean EMG amplitude; time spent at 'rest' « 6.7% RVE); and time spent at 'large activation levels' (> 100% RVE). Exposure variability due to subjects, day within subject, work cycle within day, and normalization trial was determined using a variance component analysis based on ANOV A algorithms. For all parameters, the dominate source of variance occurred between subjects. For all parameters, except the 1 st APDF percentile and the time spent at rest, the proportions of variance due to norrnalization and cycle were no more than 5% of the sum of all variance components. The betweenday variance for mean and all APDF parameters besides the 1 st APDF percentile ranged from 7 -19% of the total variance, indicating changes in work style or activation patters between days do exist. The effects of normalization reported here are likely at the low end of the range (test-retest coefficient of variation = 0.07) compared to the expected effects when data are collected in the field using less meticulously controIled RVE trials.