To develop a short measure of physical function for knee osteoarthritis (OA) using multi-national data from individuals with varying degrees of severity of knee OA.Rasch analysis, based on the partial credit model, was conducted on Knee injury and Osteoarthritis Outcome Score and Western Ontario McMaster Universities' Osteoarthritis Index data from individuals with knee OA, ranging from community to pre-total knee replacement samples from five countries. Fit of the data to the Rasch model was evaluated by overall model fit and item-level fit statistics (chi(2), size of residual, F-test). Invariance across age, gender and country was evaluated. Unidimensionality was evaluated by factor analysis of residuals. The derived short measure was further tested for fit through re-analyses in individual sub-samples. A nomogram converting raw summed scores to Rasch-derived interval scores was developed.Thirteen data sets were included (n=2145), with an age range of 26-95 years, and a male/female ratio of 1:1.4. The final model included seven of the original 22 items. From easiest to most difficult, the items (logit) were as follows: rising from bed (1.366), putting on socks/stockings (1.109), rising from sitting (0.537), bending to the floor (0.433), twisting/pivoting on injured knee (-0.861), kneeling (-1.292) and squatting (-1.292). Sub-sample analyses confirmed findings.Based on the use of accepted Rasch-based measurement methods and the compliment of countries, languages and OA severity represented in this study, our seven item short measure of physical function for knee OA is likely generalizable and widely applicable. This measure has potential for use as the function component in an OA severity scoring system.