Background & Aims: This study evaluated the construct validity of the Indonesian version of WHODAS 2.0 among persons with physical impairment.Methods: WHODAS 2.0 was self-administered to 212 participants with physical impairment in Bandung, Indonesia. The Rasch model was used to analyse the instrument’s construct validity.Results: Data of 212 participants were analysed. The average outfit mean-square (MnSq) of the 36-item and 32-item versions of WHODAS 2.0 satisfied the Rasch model expectations (0.99±0.28 vs 0.99±0.24). Item D2.5 ‘walking a long distance such as a kilometre’ and item D4.5 ‘sexual activity’ were identified as misfitting items in both versions (infit or outfit MnSq >1.4). The variance explained by measures of the 32-item version was 56.7％, and the 36-item version was 49.0％. Both versions’ reliability and separation index were excellent, with Cronbach’s alpha >0.90 and a separation index >2. The response category function and targeting did not fully satisfy the Rasch model expectations. A strong correlation between both versions of WHODAS 2.0 and WHOQOL BREF (r>0.60) established the convergent validity.Conclusions: The Indonesian version of the 32-item version of WHODAS 2.0 has acceptable construct validity in a physical impairment sample in Bandung, Indonesia.