Purpose: To examine psychometric properties of the Polish version of the 36-item WHO Disability Assessment Schedule 2.0 (WHODAS 2.0) in the population with hip and knee osteoarthritis (OA). Methods: This was a longitudinal study with repeated measures during retest examinations. Subjects from a Polish Specialist Hospital (age = 68.3 ± 9.2years, 71% female, 44.2% knee OA, 55.8% hip OA) were tested three times. They completed the Polish version of the 36-item WHODAS 2.0, the SF-36 Health Survey 2.0, the Western Ontario and Macmaster Universities Osteoarthritis Index 3.1, the Hospital Anxiety and Depression Scale, and the Numerical Rating Scale. Results: The 36-item WHODAS 2.0—Polish version demonstrated high internal consistency (Cronbach's alpha for total = 0.94), and test–retest reliability (Total ICC2,1 = 0.98). High construct validity was found as 12 out of 15 a priori hypotheses (80%) were confirmed. Most domains and Total Scores in the 36-item WHODAS 2.0 (Total ES = − 0.62, SMR = − 1.09) showed a moderate degree of responsiveness. Minimal clinically important difference (MCID) for the Total WHODAS 2.0 was 3.29 in patients undergoing rehabilitation for knee or hip OA. Conclusions: The Polish version of the 36-item WHODAS 2.0 assesses disability according to ICF in a reliable, valid and responsive way. Therefore, it provides considerable support in clinical practice and national and international scientific research of patients with hip or knee OA. [ABSTRACT FROM AUTHOR]