The present study compared three procedures for normalization of upper trapezius surface electromyographic (EMG) amplitudes: (a) a ramp procedure (providing data in per cent of maximal voluntary contraction, MVC); (b) a constant force procedure based on two reference contractions (two-force procedure) (%MVC) and (c) a procedure expressing muscle activation in per cent of a reference voluntary electrical activity (%RVE). The study also evaluated the repeatability of the ramp and the RVE procedures and estimated the force exertion (%MVC) corresponding to the RVE. To illustrate the ergonomic effect of different normalization procedures, trapezius EMG during two work tasks was compared after normalization by the two-force and the RVE procedures. Fifteen subjects participated in the whole study. We found that force estimates obtained by the ramp procedure equation could be translated to force estimates obtained by the two-force procedure by the equation: %MVC(2force) = - 0.6 + 0.9 *%MVC(ramp), although with a considerable imprecision due to large inter-individual differences. In the ramp procedure, the intra-individual test-retest coefficient of variation (CV) depended on the force level; it was 45% at 5% MVC and 10% at 30% MVC. The CV of the RVE was 15%. The reference contraction used in the RVE procedure corresponded from 13-79% MVC (median 33%MVC). The load reducing effect of an ergonomic intervention was less obvious with the RVE procedure than with the two-force procedure due to a larger inter-individual variation. The advantages and disadvantages of the different procedures are discussed.