Background and aims Endogenous pain modulation can be studied in humans by conditioned pain modulation (CPM): pain induced by a test stimulus is attenuated by a distantly applied noxious conditioning stimulus. The detection of impaired CPM in individual patients is of potential importance to understand the pathophysiology and predict outcomes. However, it requires the availability of reference values. Methods We determined reference values of CPM in 146 pain-free subjects. Pressure and electrical stimulation were the test stimuli. For electrical stimuli, we recorded both pain threshold and threshold for the nociceptive withdrawal reflex. Cold pressor test was the conditioning stimulus. The 5th, 10th and 25th percentiles for the three tests were computed by quantile regression analyses. Results The average thresholds increased after the conditioning stimulus for all three tests. However, a subset of subjects displayed a decrease in thresholds during the conditioning stimulus. This produced negative values for most of the computed percentiles. Conclusions This study determined percentile reference values of CPM that can be used to better phenotype patients for clinical and research purposes. The negative value of percentiles suggests that a slightly negative CPM effect can be observed in pain-free volunteers. Implications Pain facilitation rather than inhibition during the conditioning stimulus occurs in some pain-free volunteers and may not necessarily represent an abnormal finding.