OBJECTIVE: Body mass index (BMI) and knee osteoarthritis have a strong association, but other anthropometric measures lack such associations. To date, no study has evaluated non-obese knee osteoarthritis to negate the systemic and metabolic effects of obesity. This study examines the validity of the contention that BMI and other anthropometric measures have a significant relationship with knee osteoarthritis. METHODS: In total, 180 subjects with a diagnosis of knee osteoarthritis were recruited and classified according to Kellgren-Lawrence (KL) grades. Body mass index, mid-upper arm circumference, waist-hip ratio and triceps-skinfold thickness were recorded by standard procedures. Osteoarthritis outcome scores (WOMAC) were evaluated. RESULTS: (1) In both genders, the BMI was significantly higher for KL grade 4 than for grade 2; triceps-skinfold thickness was positively correlated with the joint space width of the tibial medial compartment. (2) In males, triceps-skinfold thickness significantly increased as the KL grades moved from 2 to 4; the significantly higher BMI found in varus aligned knees was positively correlated with WOMAC scores. (3) In females, the waist-hip ratio was significantly higher for KL grade 4 than for grade 2; a significant correlation was found between BMI and WOMAC scores. The waist-hip ratio was significantly associated with varus aligned knees and it positively correlated with WOMAC scores and with the joint space width of the tibial medial compartment. The mid-upper arm circumference demonstrated no correlation with knee osteoarthritis. CONCLUS'ON: This study validates the contention that BMI and other anthropometric measures have a significant association with knee osteoarthritis. Contrary to common belief, the triceps-skinfold thickness (peripheral fat) in males and the waist-hip ratio (central fat) in females were more strongly associated with knee osteoarthritis than BMI.