The Kellgren and Lawrence (K/L) system and Osteoarthritis Research Society International (OARSI) atlas are frequently used to define radiographic knee osteoarthritis (OA). The purpose of the current study was to determine the extent to which tibiofemoral OA rates differ between the K/L system and OARSI atlas criteria and to compare qualitative (K/L and OARSI) and quantitative (millimetres) measures of joint space narrowing (JSN).Posteroanterior radiographs of 1,178 knees, from 621 individuals with varying severity of OA, were graded by a trained physician with the K/L system (grade 0-4) and the OARSI atlas (osteophytes/JSN graded 0-3). Using the K/L system, the presence of OA was defined with the traditional cut-off of ≥grade 2 (definite osteophyte and possible JSN) and an alternative cut-off of at least a definite osteophyte alone (≥grade 2/osteophyte). For the OARSI atlas, OA was considered present if the sum of osteophytes or JSN ≥grade 2, or grade 1 JSN occurred in combination with grade 1 osteophyte. Minimum joint space width (mJSW) was measured manually in millimetres.According to the K/L system (≥grade 2), 167 knees (14.2 %) had tibiofemoral OA and 203 (17.3 %) had ≥grade 2/osteophyte. In contrast, 309 knees (26.2 %) had tibiofemoral OA according to OARSI atlas criteria. K/L and OARSI JSN descriptions were significantly associated with mJSW (p < 0.022).Radiographic tibiofemoral OA was almost twice as common using OARSI atlas criteria compared with the K/L system. This discrepancy is likely to contribute to the large variability of OA prevalence observed in the literature and is important for clinicians to consider when diagnosing radiographic OA. The cut-off for defining radiographic knee OA using the two systems should not be considered comparable.III.