Purpose: Although correlates of complementary and alternative medicine (CAM) have been described using cross-sectional study designs, data describing use patterns over time are sparse. The purpose of this study was to describe correlates of CAM usage patterns among persons with radiographically confirmed knee osteoarthritis (OA). Methods:We included 2,114 participants of the Osteoarthritis Initiative with radiographic tibiofemoral knee OA in at least one knee at baseline who had five assessments completed over 4 years. Trained interviewers asked “During the past 6 months, did you use the following health supplements for joint pain or arthritis?” with separate questions for chondroitin sulfate and glucosamine (GLU/CHON). Participants also reported use of provider based CAM (e.g. acupuncture, ayurveda, naturopathy, biofeedback, homeopathy, Reiki, chiropractic, message) and patient self-directed CAM therapies (e.g. biologically-based supplements (SAME, MSM, vitamins, herbs), biologically-based diets, biologically-based topical agents, magnet therapy, mind-body interventions (Tai Chi, Yoga, Chi Gong, or Pilates, relaxation, spiritual activities)). Correlates of treatment approach for OA considered included sociodemographic indicators, body mass index, overall measures of mental and physical well-being, and clinical indices of knee OA. Polytomous logistic regression provided adjusted odds ratio estimates (aOR) and 95% confidence intervals (CI). Results: Fifty-five percent reported no use of GLU/CHON at any of the five assessments, 18.6% reported use at one or two times, 13.4% at three or four times, and 12.8% at all assessments. Relative to non-Hispanic Whites, Blacks had reduced odds of reports of GLU/CHON use on multiple assessments (aOR use 1-2 times: 0.59; (95% CI: 0.41-0.86); aOR use 3-4 times: 0.38; (95% CI: 0.23-0.61); aOR use 5 times: 0.13; (95% CI: 0.060.28)). Those with greater than college graduate education relative to those with < high school education had increased odds of GLU/CHON use on multiple assessments (aOR use 1-2 times: 1.55; (95% CI: 1.03-2.32); aOR use 3-4 times: 1.91; (95% CI: 1.16-3.16); aOR use 5 times: 1.69; (95% CI: 1.02-2.80)). Worst pain at baseline was inversely associated with GLU/ CHON use on every assessment (WOMAC Pain aOR use 5 times: 0.69; (95% CI: 0.53-0.89). Increased quality of life at baseline was inversely correlated with GLU/CHON use at every assessment ((KOOS Quality of Life aORuse 5 times: 0.81; (95% CI: 0.65-1.00). Increased K-L grade tended to correlate with increased GLU/CHON use reports on multiple assessments. Patients reporting other self-directed CAM practices at baseline were more likely to report GLU/CHON use on multiple assessments ((aOR use 1-2 times:1.94; (95% CI: 1.46-2.57); aOR use 3-4 times: 2.36; (95% CI:1.71-3.26); aOR use 5 times: 3.15; (95% CI: 2.26-4.37)). Relative to Whites, Blacks were more likely to report self-directed CAM use on multiple assessments and less likely to report provider-based CAM use. Women were more likely to report self-directed CAM use on multiple assessments. Patients who reported multiple CAM approaches at baseline (e.g. patient-based CAM, provider-based CAM, GLU/CHON) were more likely to report use of such approaches on multiple assessments. Conclusions: CAM therapies are commonly used to treat joint and arthritis pain among persons with knee OA, but reports of use over time is limited. The extent to which low use of CAM on multiple assessments is owing to lack of effectiveness or other factors remains unknown.