Seventeen patients with primary osteoarthritis of the knee were evaluated with respect to the severity and clinical importance of pain, stiffness and physical function during the conduct of a double-blind randomized controlled trial of flurbiprofen SR versus diclofenac sodium SR using the WOMAC Osteoarthritis Index. Mean importance scores were similar for items within the same dimension as well as between items in different dimensions. In general, low levels of correlation were noted between the severity and importance of symptoms. Analysis of individual WOMAC items within a given subscale indicated that, although highly correlated, they differed from one another. Factor analysis further supported the contention that scores from items within a subscale could be summated into subscale scores. These observations are of importance in the weighting and aggregation of items within discrete dimensions and have the potential for reducing sample size requirements for clinical trials in osteoarthritis.