Objective: Osteoarthritis is acknowledged as an enduring condition, however, in epidemiological studies, half the participants who report having osteoarthritis at one time report not having it at a subsequent time. The aim of this study was to examine whether variations in reporting doctor-diagnosed osteoarthritis reflected concurrent fluctuations in indicators of disease severity in mid-age women. Methods: Data were from 7,623 7,589 participants (aged 50-55 years in 2001) in the Australian Longitudinal Study on Women’s Health. Based on self-report of doctor-diagnosed osteoarthritis at surveys in 2001, 2004, 2007 and 2010, the participants were classified according to pattern of osteoarthritis reporting (e.g. 0-0-0-0=‘no’ on all surveys, 0-1-0-1=‘no-yes-no-yes’). Indicators of disease severity included use of anti-inflammatory medications, frequency of joint pain/stiffness, use of anti-inflammatory medications, and physical functioning assessed with the SF-36. Bar graphs were used to show concurrent variations in osteoarthritis and markers, and associations were examined using log-linear models. Results: In this sample, 46% reported having osteoarthritis on at least one survey, with half these cases reporting not having osteoarthritis at a later survey. Odds of reporting joint pain/stiffness often (odds ratio (OR) 7.26, 95% confidence interval (CI) 7.06-7.47) and using anti-inflammatory drugs (OR 4.44, CI 2.37-8.33) were higher, and physical functioning scores were lower (OR 3.75, CI 3.56-3.95) when participants reported having osteoarthritis. Conclusion: Variations in reporting osteoarthritis coincided with episodic fluctuations in symptoms and functioning. Inconsistent reporting of osteoarthritis could therefore seems to reflect presence of symptoms rather than measurement error and should be considered in longitudinal studies.