OBJECTIVE: Serum hyaluronan (HA) has been proposed as a potential biomarker of osteoarthritis (OA). We examined associations between serum HA and radiographic OA in an ethnically diverse, population-based sample. METHODS: Participants were selected from the Johnston County Osteoarthritis Project, using stratified simple random sampling to achieve balance according to radiographic knee OA status, ethnic group, sex, and age group. Serum HA was measured by enzyme-linked immunosorbent assay. Radiographic OA variables included knee OA, knee OA laterality, knee OA severity, concomitant knee and hip OA, and total number of OA-affected knee and hip joints. Analysis of covariance was used to assess differences in mean serum levels of natural log-transformed HA (ln serum HA) between groups, adjusting for ethnicity, sex, age, body mass index (BMI), and self-reported comorbidities. RESULTS: Levels of ln serum HA were positively associated with all definitions of radiographic OA (P < 0.0001). Levels of ln serum HA were higher in Caucasians (P = 0.0094) and in men (P = 0.0038) and were moderately correlated with age (r = 0.35, P < 0.0001). The associations with radiographic OA, ethnicity, sex, and age remained statistically significant after adjustment (P < 0.0045). There were no interactions between ethnicity and the other covariates. CONCLUSION: These cross-sectional data support a role for serum HA as a biomarker of radiographic OA. The variations in levels of serum HA attributable to ethnicity, sex, and age were not explained by radiographic OA, BMI, or comorbidities. The lack of strong confounding between serum HA and comorbidities further supports a role for serum HA as a potential biomarker.