There are many existing architectural structures that need reinforcement because of low construction quality, absurd design, inappropriate use, accidental natural disaster and building function change etc. Since there are several possible reinforcement schemes available to one structure, choosing the optimal scheme becomes a crucial problem that a designer must face. The optimization of reinforcement schemes needs considering not only quantitative factors like construction cost and period, but also qualitative index, which is difficult to quantize, such as durability and construction difficulty degree. Therefore, scheme selection is a representative multi-index semi-structural problem. This poses considerable difficulty to designers and is complicated to regulate because of the subjective randomness. In this paper, fuzzy multi-index half-structural theory is applied to the selection of building reinforcement schemes. By determining relative membership degree matrix and objective weight matrix of each index, superior degree of each scheme to decision is decided; objective index and designers’ experience are combined effectively, and then optimal reinforcement scheme can be obtained. The results reveal that the calculate results are well consistent with those of test, and have great computational accuracy.