The purpose of this paper was to compare the effects of different data reduction procedures on the values of variables characterizing the time pattern of trapezius muscle activity during full work shifts. Surface electromyography (EMG) of the right and left upper trapezius muscles were obtained from 40 young subjects in different occupations, mainly electricians, hairdressers and students. The target EMG variables were gap frequency, muscle rest, and the number and duration of episodes with sustained muscle activity (from 0.13 s to 30 min as minimum duration). These variables were derived from the EMG recordings using different Root Mean Square (RMS) windows (from 0.13 to 6.38 s), and discrimination levels between "activity" and "rest" (0.5, 1 and 2 % of maximal EMG). The results give basis for practical suggestions for EMG analyses of full work shifts. For most variables, a discrimination level of 0.5% EMGmax showed to be preferable. The time proportion of muscle rest and sustained muscle activity should, in general, be preferred over the corresponding frequency measures. Sustained muscle activity should be calculated using a RMS window between 1 and 3 s, and preferably be stated in terms of variables describing time proportions of activity. Uninterrupted activity episodes longer than 10 min proved not to be a useful variable due to limited occurrence in many work shifts.