Background: A set of physical tests representing typical activities relevant to persons with knee osteoarthritis (KOA) is recommended by the Osteoarthritis Research Society International (OARSI). The objective of this study was to designate the cut-off values for OARSI-recommended performance-based tests to better discriminate disease severity in patients with KOA. Methods: As recommended by the OARSI, we conducted the 30-s chair stand test (30-s CST), 40-m fast-paced walk test (40-m FPWT), stair climb test (SCT), 6-min walk test (6MWT), and timed up and go test (TUG). To investigate the discriminative power of the performance test scores using the Kellgren–Lawrence (K-L) grading system, receiver operating characteristic (ROC) curve analysis was performed. Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) was used to assess self-administered health status. Results: Forty-eight patients with bilateral KOA were included this study. The area under the curves for four of the five performance tests (30-s CST, six-step SCT, 40-m FPWT, and TUG) had acceptable discrimination (0.753–0.793), while the 6MWT had excellent discrimination (0.860). The stepwise multiple regression analysis demonstrated that 6MWT and 30-s CST were significant and independent determinants of WOMAC score, explaining 48% of the variance (P < 0.001). Conclusion: The OARSI performance tests can discriminate disease severity in individuals with KOA. The 6MWT has the highest level of discriminative validity among all. In addition, despite the correlation between all five tests and the WOMAC score, 6MWT and 30-s CST are independent and significant determinants of the WOMAC.