BACKGROUND: Epidemiological studies have, so far, identified factors associated with increased risk for incident or progressive OA, such as age, sex, heredity, obesity, and joint injury. There is, however, a paucity of long-term data that provide information on the nature of disease progression on either group or individual levels. Such information is needed for identification of study cohorts and planning of clinical trials. The aim of the study was, thus, to assess the variation in pain and function on group and individual level over 2 years in previously meniscectomized individuals with and without radiographic knee osteoarthritis (OA). METHODS: 143 individuals (16% women, mean age at first assessment 50 years [range 27-83]) were assessed twice; approximately 14 and 16 years after isolated meniscectomy, with a median interval of 2.3 years (range 2.3-3.0). Radiographic OA (as assessed at the time of second evaluation) was present in the operated knee in 40%, and an additional 19% had a single osteophyte grade 1 in one or both of the tibiofemoral compartments. Subjects completed the self-administered and disease-specific Knee injury and Osteoarthritis Outcome Score (KOOS). RESULTS: There were no significant changes in the group mean KOOS subscale scores over the 2-year period. However, a great variability over time was seen within individual subjects. Out of 143 subjects, 16% improved and 12% deteriorated in the subscale Pain, and 13% improved and 14% deteriorated in the subscale ADL > or = 10 points (the suggested threshold for minimal perceptible clinical change). Similar results were seen for remaining subscales. CONCLUSION: Group mean scores for this study cohort enriched in incipient and early-stage knee OA were similar over 2 years, but pain, function and quality of life changed considerably in individuals. These results may be valid also for other at risk groups with knee OA, and motivate further careful examination of the natural history of OA, as well as properties of the OA outcome instruments used. Longitudinal outcome data in OA studies need to be analyzed both on an individual and a group level.