Abilities of the Short-Form McGill Pain Questionnaire to assess change have scarcely been addressed in previous studies. The aim of the present study was to examine test–retest reliability, sensitivity to change and responsiveness to clinically important change using a Norwegian version (NSF-MPQ) in different groups of patients. ICC(1,1) values for test–retest reliability (relative reliability) assessed 1–3 days apart for total, sensory and affective scores were, respectively, 0.75, 0.76 and 0.62 in patients with musculoskeletal pain (n=58), and 0.93, 0.95 and 0.79 in patients with rheumatic pain (n=25). Variability in total scores (absolute reliability) was less in patients with rheumatic pain (within-subject standard deviation, Sw=2.70) than in patients with musculoskeletal pain (Sw=4.28). Sensitivity to change by standardized response mean (SRM) was mostly large (>0.80) for three patient groups reporting improvement after treatment. More sensitivity to change was demonstrated by the total and sensory scores than by the affective score, and sensitivity of the total score was similarly good to capture improvement as the Visual Analogue Scale (VAS). Indication was provided that mean improvement of groups in NSF-MPQ total scores should be >5 on the 0–45 scale to demonstrate a clinically important change. Responsiveness to clinically important change by receiver operating characteristic curve analysis was modest, as area under the curve indicating ability to discriminate improved and not improved patients with musculoskeletal pain, was only 0.61. The study indicates mostly satisfactory test–retest reliability and responsiveness values of the NSF-MPQ, but shows that the measurement properties vary between groups of patients with pain.