Little is known about the measurement properties of numeric rating scales (NRS) for pain in AD. We evaluated a novel NRS for skin-pain and existing NRS for average overall-pain in adults with AD. Self-administered questionnaires and skin-examination were performed in 463 AD patients (age 18-97 years) in a dermatology practice setting. Numeric rating scales skin-pain and average overall-pain had moderate correlations with each other, and multiple clinician-reported and patient-reported AD severity outcomes (Spearman correlations, P < 0.0001). There were significant and stepwise increases of NRS skin-pain and average overall-pain scores with patient-reported global severity (Wilcoxon rank-sum test, P < 0.0001). Floor-effects were observed for NRS skin-pain and average overall-pain. Changes from baseline in NRS skin-pain and average overall-pain showed weak-moderate correlations with changes of POEM, vIGA-AD*BSA, SCORAD, and DLQI. Using an anchoring approach, the optimal interpretability band for NRS skin-pain was clear = 0, mild = 1-3, moderate = 5-6, severe = 7-9, and very severe = 10 (weighted kappa = 0.4923). The thresholds for minimally clinically important difference for NRS skin-pain ranged from 2.2 to 2.9. NRS skin-pain and average overall-pain showed moderate-good reliability. Numeric rating scales skin-pain and average overall-pain had sufficient validity, reliability, responsiveness, and interpretability in adults with AD, and were inherently feasible as single-items for use in clinical trials and practice.