The strength of the associations of knee injury and obesity with osteoarthritis of the knee was studied for 3,885 adults aged 45–74 years who received knot x-rays in the First National Health and Nutrition Examination Survey, 1971–1975 Bilateral osteoarthritis was more prevalent (5%) than unilateral osteoarthritis (2%). Bilateral osteoarthritis was twice as prevalent in women as in men; however there was no sex difference in the prevalence of unilateral osteoarthritis. Odds ratios calculated by means of polychotomous logistic regression indicated thai obesity, knee injury, and age were significantly associated with both unilatera and bilateral knee osteoarthritis. Obesity was a stronger predictor of bilatera osteoarthritis than was knee injury (odds ratio (OR)=6.6 for obesity, 3.5 for right knee injury, and 3.0 for left knee injury; 95% confidence interval (Cl) 4.71-9.18 1.80-6.83, and 1.51-6.11, respectively). Knee injury was a stronger predictor ol unilateral osteoarthritis than was obesity (OR=3.4 and 2.4 for obesity in the right and left knee, respectively (95% Cl 1.55-7.29 and 0.96-5.75) and OR=16.3 and 10.9 for injury in the right and left knee, respectively (95% Cl 6.50-40.89 and 3.72-31.93)). These findings suggest that different pathogenetic processes may exist for unilateral and bilateral knee osteoarthritis.