OBJECTIVES: Maximum Difference Scaling (MaxDiff) is a survey research technique for obtaining preference scores for a set of items that provides greater discrimination between items while indicating how strongly an item is preferred. This study investigated the use of MaxDiff as a means of determining patient-reported importance of medication treatment attributes. METHODS: MaxDiff was used in a survey of rheumatoid arthritis (RA) patients to determine the importance and relative rank of specific attributes of RA treatments. The following attributes were selected based upon literature review and opinions of the research team: 1) Reduces pain; 2) Potential side effects; 3) How often treatment taken; 4) How treatment is given; 5) Where treatment is given; 6) Personal costs; 7) Works quickly; 8) How long treatment effects last; 9) Keeps disease from getting worse; and 10) Improves physical abilities. Respondents were shown 10 sets of 4 attributes and, for each set, were asked to indicate the RA treatment attribute that was most important and least important to them. The attribute sets were selected using an experimental design that showed each attribute an equal number of times and in different order within the sets. Hierarchical Bayesian techniques were used to derive respondent-level attribute importance scores and, based on the importance scores a relative rank order was developed for each attribute.   RESULTS: Based on 291 surveys, MaxDiff attribute importance scores ranged from 13 to 209 with higher scores indicating increased importance. Preventing the disease from getting worse, improving physical abilities, and reduction in pain were the most important RA treatment attributes, all having scores >195, while the ‘how’, ‘when’, and ‘where’ treatment administration attributes were the least important, all having scores ­<15. CONCLUSIONS: MaxDiff importance scores demonstrated discrimination among attributes and respondents and should be considered as an alternative to more traditional ranking and rating approaches.