Due to inter-operator variability, two operators were used to assess the consistency of motor unit (MU) identification during ramp contractions, by the comparison of semi-automatic decompositions of the same recordings. Static shoulder abduction was performed against a force transducer in a position with the upper arms vertical and elbows flexed to 90 degrees. The subjects followed an 8-s force trajectory: 30% maximum voluntary contraction (MVC, 2 s), a reduction in force from 30% to 0% MVC (2 s), 0% MVC (1 s), an increase in force from 0 to 30% MVC (2 s), and 30% MVC (1 s). Muscle activity was recorded from the supraspinatus muscle with a quadripolar needle. From six recordings of 8 s duration, a total of 2527 MU firings were identified by both operators, and 93% of these were identified identically into 31 MUs. Both operators identified 8 of these MUs as continuously firing, 5 as only being active either before or after the 1 s at 0% MVC, and 18 as being de-recruited during force decreases and recruited during force increases. Both operators agreed that 16 of these 18 MUs were de-recruited at a higher force level than that at which they were recruited, which may be due to the electromechanical delay. The coefficient of variation for double determination of the results obtained by operators A and B was 8.5% for the number of MU firings, 4.5% for the MU mean firing rate, and 8.4% for the MU action potential (MUAP) amplitude. Therefore, the operator interactive decomposition method was considered to be valid for studying recruitment and de-recruitment as well as firing rate and MUAP amplitude during static, force-varying ramp contractions.