ObjectiveTo assess the diagnostic performance of a comprehensive set of tests and measures to discriminate patellofemoral (PF) from tibiofemoral (TF) osteoarthritis (OA).MethodsThe Clinical Assessment of the Knee Study is a study of knee pain in the general population. The presence of PF crepitus and pain with PF compression were assessed. Anterior knee pain was determined using a knee pain map. Pain with climbing/descending stairs and walking on level ground were assessed with the Western Ontario and McMaster Universities Osteoarthritis Index. Radiographs were used to define the compartment(s) of the knee affected by OA (no OA, isolated/predominant PF joint OA, and isolated/predominant TF joint OA). In knees with mixed OA, knees with more severe PF joint OA were included in the isolated/predominant PF group (the same was done for TF joint OA). We determined the sensitivity, specificity, positive predictive value (PPV), negative predictive value, and positive likelihood ratio (LR+) for each test and measure individually, and the combination of these measures, in distinguishing knees with PF joint OA from knees with TF joint OA or no OA.ResultsThere were 745 knees included in the study. No measure had high sensitivity and specificity. Pain with climbing/descending stairs had the greatest sensitivity (90%) but poor specificity (15%). The combination of definite crepitus with no pain on walking had the greatest specificity (96%), PPV (53%), and LR+ (1.8), but poor sensitivity (7%).ConclusionTypical clinical examination findings and knee pain patterns commonly thought to represent underlying PF joint pathology do not discriminate knees with PF joint OA from knees without OA or TF joint OA.