Clinical observations and recent studies suggest that arthritis patients vary considerably in their involvement in self-management efforts. In the literature on health promotion, there is growing recognition that patients may be at different stages of change with respect to the adoption of self-management strategies. The major goal of the present study was to examine whether cluster analysis could be used to identify homogeneous subgroups of patients having persistent arthritis pain based on their responses to a stages of change questionnaire. Participants in this study (103 patients having rheumatoid arthritis and 74 patients having osteoarthritis) completed a stages-of-change measure specific to adoption of a self-management approach to their arthritis. A cluster analysis identified five distinct subgroups of arthritis patients: (1) precontemplation - 44% of the sample; (2) contemplation - 11% of the sample; (3) preparation - 22% of the sample; (4) unprepared action - 6% of the sample; and (5) prepared maintenance - 17% of the sample. These subgroups are generally consistent with what might be expected based on the transtheoretical model of stages of change by Prochaska and DiClemente (Prochaska JO, DiClemente CC. Towards a comprehensive, transtheoretical model of change: states of change and addictive behaviors. In: Miller WR, Heather N, editors. Applied clinical psychology, 2nd ed. Treating addictive behaviors, New York: Plenum Press, 1998. pp. 3-24.), and may have important clinical implications. For example, it is possible that the arthritis subgroups identified may predict arthritis patients' participation in and responsiveness to pain-coping skills training, exercise interventions, or other formal self-management training programs. Also, one may be able enhance the outcomes of self-management interventions for arthritis by tailoring treatment to the patient's particular stage.