The relative effects of repetition, force and posture were studied in order to investigate how continuous biomechanical measurements can be combined into a single metric corresponding to subjective discomfort. A full factorial experiment was conducted involving repetitive wrist flexion from a neutral posture to a given angle against a controlled force. Seven subjects performed the task using two paces (20 and 4 motions/min), two force levels (15 and 45 N) and two angles (15 and 45 degrees) for 1 h each. Discomfort was reported on a 10 cm visual analogue scale anchored between 'no discomfort' and 'very high discomfort'. Repeated measures analysis of variance showed that all main effects were statistically significant (p < 0.05) and no significant interactions were observed. A linear regression model was fitted to the data and used for generating frequency weighted digital filters that shape continuous recordings of repetitive motions and exertions into an output proportional to relative discomfort. The resulting high-pass digital filter had a 22 dB/decade attenuation slope. A simulated industrial task used for validating the model involved repetitively transferring pegs across a horizontal bar and inserting them into holes against a controlled resistance. Angular wrist data were recorded using an electrogoniometer and filtered. Six subjects performed the task of the three conditions consisting of (1) 15 wrist flexion, 15 N resistance and 6 motions min, (2) 15 wrist flexion. 45 N resistance and 12 motions/min, and (3) 45 degrees wrist flexion, 45 N resistance and 15 motions/min. Subjective discomfort was reported after performing the task for 1 h. Pearson correlations between subjective discomfort ratings and the integrated filtered biomechanical data for individual subjects ranged from 0.90 to 1.00. The pooled correlation across subjects was 0.67. This approach may be useful for physical stress exposure assessment and for design of tasks involving repetitive motions and exertions.