The use of percentage pain reduction is increasingly used to evaluate the effectiveness of pain treatments, but the degree of agreement between calculated percentage pain reduction (CPPR) as calculated from pre- and post-treatment levels of pain intensity and those reported directly by patients is unknown. Lack of agreement between these two measures could lead to errors in the determination of treatment effectiveness. We aimed to determine the agreement between CPPR and patient-reported percentage pain reduction (PRPPR). Patients with acute or cancer pain were asked to rate their pain intensity on a 0-10 verbal numerical rating scale (NRS) and to estimate the percent pain reduction from baseline pain after analgesic administration. They then received analgesics every 10 min until pain intensity declined to 4/10 or less. To evaluate agreement between CPPR and PRPPR, we computed the concordance correlation coefficient (CCC), which measures both accuracy and precision, and estimated the 95% limits of agreement for the differences between these two measures. 761 adult patients were enrolled. Female, healthy patients with acute pain of severe intensity and high levels of education predominated in the sample. The mean difference between CPPR and PRPPR was -2.6% (95% limits of agreement -12 to 17%). The CCC was 0.56 (accuracy was 0.9 and precision was 0.6). Although CPPR appeared to underestimate PRPPR in the higher range, this trend was not clinically important. The agreement between percentage pain reductions calculated from NRS scores and those estimated by patients did not vary according to gender or age. The good overall agreement between percentage pain reductions calculated from NRS scores and those estimated by patients suggests that these indices may be used interchangeably. The findings of this study extend existing patient-centered pain research and may be applied for the evaluation and comparison of pain treatments.