Objective To develop the short version of the elderly disability assessment scale (EDAS-SF), and to evaluate its validity and reliability. Methods We conducted a face-to-face investigation in five rural communities in Suining and three urban communities in Chengdu city for elderly adults. We consulted gerontologists and rehabilitation physicians through an internet survey. Proper items were selected from the EDAS according to the results from the following statistical methods: critical ration, Cronbach’s α, correlation coefficient and exploratory factor analysis. Based on these work, the EDAS-SF was developed. The Cronbach’s α was used to assess the internal reliability and confirmatory factor analysis (CFA) to assess the construct validity of the EDAS-SF. Results A totally of 916 elderly adults were surveyed, among which, 887 finished the questionnaire, accounting for 96.8%. The mean age of the studied population was 70.8±6.9 years old. Finally, seven items were chosen for the EDAS-SF which included mental function, organ function, communication, activity, self-care, family function, economic and social function. The Cronbach’s α of the scale was 0.836. Except for the " economy and social function” dimension and the " communication” dimension, the Cronbach’s α of other dimensions were higher than 0.8. EDAS-SF was proved to have good internal reliability. In addition, the CFA indicated that the theoretical model of the EDAS-SF was well fit to the data of the checking sample, which meant EDAS-SF had good construct validity. Conclusion The current study develops the short version of elderly disability assessment scale. The internal reliability and construct validity of the EDAS-SF are good.