This research proposes an input modality decision model to reinforce universal accessibility for the elderly in smart environments which have high interaction complexity and contains diverse smart devices. To do this, following tasks were conducted: 1) elderly-oriented disability was defined and classified as permanent physical disabilities and contextual temporal disabilities. 2) Knowledge bases of input modality decision model were redesigned, which are consisted of multi-layered relationships between user, environment and device context. Based on this, 3) relationship priority was analyzed using QFD, and 4) temporal disability expressed in heterogeneous and ambiguous manner was updated and reflected to decide an adaptive input modality through fuzzy logic. The results of this research enable to suggest an elderly-personalized accessibility which reinforces input modality. In addition, this model can recognize and update dynamic temporal disabilities and confliction between activated input modality and environmental context. Based on this research, when the proposed model is applied to the smart control system, environment utilization ability gap between general users and the elderly can be reduced, and the basis of an interaction architecture in a smart environment can be built.