OBJECTIVE: To compare three questions on knee pain with respect to determined prevalence and associations with disability and structural change. METHODS: Postal survey to 4057 men and women aged 40-79 years. Knee pain was defined by three questions: (A) "Have you ever had pain in or around the knee on most days for at least a month? If so, have you experienced any pain during the last year?" (B) "Have you had pain within the last year in or around the knee that occurred on most days for at least a month?" (C) "Have you had knee pain on most days of the last month?" [American College of Rheumatology (ACR) criteria for knee osteoarthritis]. Disability was assessed with the SF-36 health status questionnaire. Radiographs (AP weight bearing and skyline) were obtained on a proportion (n = 459) and graded for maximum osteophyte in any compartment. RESULTS: Prevalence of knee pain for questions A, B, and C were 28.3%, 25.3%, and 19.3% respectively. Highest rates of disability were observed for question C (71.3% compared with 60.9% for question A). There was no major difference between questions in terms of percentage with > or = grade 1 osteophyte or > or = grade 2 osteophyte. Sensitivity and specificity of each question for > or = grade 1 osteophyte did vary, with question A being most sensitive but least specific (58.7% and 59.1%) and question C most specific (72.7%) but least sensitive (45.4%). CONCLUSIONS: Estimates of knee pain and disability are influenced by even minor changes in question content. The ACR criteria question may be a better predictor of disability but is relatively insensitive for use in the community.