Abstract Background Patient-reported outcome measures (PROMs) are tools to screen a population, to monitor the subjective progress of a therapy, to enable patient-centred care and to evaluate the quality of care. The QUALITOUCH Activity Index (AI) is such a tool, used in physiotherapy. This study aimed to provide reference values for expected AI outcomes. Methods A large data set uniting clinical routine data and AI outcomes was generated; it consisted of data of 11,948 patients. For four defined diagnoses, i.e. chronic lower back pain, tibia posterior syndrome, knee joint osteoarthritis and shoulder impingement, the AI responses related to the dimensions “maximum pain level” and “household activity” were analyzed. Reference corridors for expected AI outcomes were derived as linear trend lines representing the mean, 1st and 3rd quartile. Results Reference corridors for expected AI outcomes are provided. For chronic lower back pain, for example, the corridor indicates that the initial average AI value related to maximum pain of 49.3 ± 23.8 points on a visual analogue scale (VAS multiplied by factor 10) should be improved by a therapeutic intervention to 36.9 ± 23.8 points on a first follow-up after four weeks. Conclusions For four exemplary diagnoses and two dimensions of the AI, one related to pain and one related to limitations in daily activities, reference corridors of expected therapeutic progress were established. These reference corridors can be used to compare an individual performance of a patient with the expected progress derived from a large data sample. Data-based monitoring of therapeutic success can assist in different aspects of planning and managing a therapy.