Although progress has been made in the clinical metrology of pain in osteoarthritis, much further work remains. The preferred methods of measurement remain debatable. In this longitudinal, open study, a comparison of eight self-rating pain scales has been conducted. A total of 333 patients entered the four-week study after completing a 3-7 day NSAID-free washout period. Patients were assigned to treatment with oxaprozin 1200 mg p.o. once daily with titration permitted between 600 mg and 1800 mg. Rescue analgesia with acetaminophen (paracetamol) 325 mg (maximum 2600 mg) was allowed. At the end of the washout and the treatment period, patients completed eight self-administered pain scales. All pain measures detected clinically important and statistically significant improvements in pain. The pain scales differed in their degree of responsiveness. The Likert and visual analogue scales and their primary variations (continuous chromatic analogue and numerical scales) were more responsive than more complex measures. A positive correlation between initial pain rating and subsequent pain relief was confirmed in this study. We conclude that, while pain is a subjective sensory phenomenon, its perceived severity can be evaluated using a variety of self-administered pain scales, all of which are capable of detecting improvements in health status following effective pharmacological intervention.