OBJECTIVE:To examine the longitudinal relationship of knee pain, radiographic osteoarthritis (OA), symptomatic knee OA, and knee pain severity to incident widespread pain. METHODS:The Multicenter Osteoarthritis Study is a longitudinal cohort of persons with or at risk of knee OA. Participants were characterized as having consistent frequent knee pain, radiographic OA (Kellgren/Lawrence scale grade ≥2), symptomatic OA, and knee pain severity at the 60-month visit (baseline). Widespread pain was categorized as pain above and below the waist, on both sides of the body and axially, using a standard homunculus, excluding knee pain. Incident widespread pain was defined as the presence of widespread pain at 84 months in those who were free of widespread pain at baseline. We assessed the relationship of baseline radiographic OA, symptomatic OA, consistent frequent knee pain, and knee pain severity, respectively, with incident widespread pain using logistic regression, adjusting for potential confounders, including models with and without pain severity. RESULTS:At baseline, 1,129 subjects were eligible for analysis (mean ± SD age 66.7 ± 7.8 years; mean ± SD body mass index 30.1 ± 5.8 kg/m2 ; 52% women). Radiographic OA in either knee (adjusted odds ratio [ORadj ] 0.90 [95% confidence interval (95% CI) 0.63-1.30]; P = 0.587) was not associated with incident widespread pain. Baseline bilateral consistent frequent knee pain (ORadj 2.35 [95% CI 1.37-4.03]), bilateral symptomatic OA (ORadj 2.11 [95% CI 1.04-4.24]), and knee pain severity (worst knee) (ORadj 1.11 [95% CI 1.05-1.17]; P < 0.001) were significantly associated with incident widespread pain. CONCLUSION:Consistent frequent knee pain, symptomatic OA, and knee pain severity increased the risk of developing widespread pain, independently of structural pathology. These results suggest that knee pain, and not structural pathology, contributes to the onset of widespread pain.