Nociceptive stimuli consisting of short pulses of gaseous CO(2) applied to the nasal mucosa were investigated in 12 healthy volunteers (six males, six females) with respect to trial-to-trial variability. A statistical rather than a psychophysical approach was employed, focusing on test-retest reliability. According to a randomized, volunteer-blind crossover design, two repeated measurements for each of four stimulus durations (200, 400, 800 and 1600 ms) were performed on eight different days. During measurements, a total of 80 stimuli of 16 concentrations of CO(2) (40-70% v/v in 2% steps) was applied with an interstimulus interval at 30 s. The interday variability of the pain ratings was lowest at a stimulus duration of 400 ms and highest at 200 ms. The results show that stimulus duration may affect the statistical outcome of studies on pain, and should be considered when setting up an experimental pain model.