This longitudinal study aimed to identify risk factors for the incidence and progression of radiographic knee osteoarthritis (OA). We examined the inhabitants of Miyagawa village aged ≥ 65 years every two years between 1997 and 2007. Anteroposterior radiographs of both knees were graded for OA using the Kellgren-Lawrence (K/L) grading system. Knee OA was defined as grade ≥ 2. We recorded the incidence of knee OA among participants in whom both knees changed from K/L grades 0 or 1 to ≥ 2 over a four-year follow-up period. We also recorded the progression of knee OA using this threshold among patients in whom one or both knees changed from K/L grades 2 or 3 to any higher grade over the follow-up period. Baseline data obtained from standard questionnaires, physical findings and X-rays included age, gender, body mass index (BMI), osteoporosis, Heberden's nodes, knee range of motion (ROM), knee pain and cigarette smoking. The rates of incidence and progression of knee OA among 360 participants (241 women, 119 men) who fulfilled the study criteria were 4.0 and 6.0% per year, respectively. Female gender (odds ratio [OR] 2.849, 95% confidence interval [CI] 1.170-6.944) and high BMI (OR 1.243, 95% CI 1.095-1.411) were significantly associated with the incidence of knee OA, and restricted knee ROM (OR 0.941, 95% CI 0.892-0.992) was significantly associated with knee OA progression. Patients with a low knee ROM relative to grade of radiographic knee OA require more careful follow-up than those with a higher ROM.