Lossy transmission is a common problem suffered from monitoring systems based on wireless sensors. Though extensive works have been done to enhance the reliability of data communication in computer networks, few of the existing methods are well tailored for the wireless sensors for structural health monitoring (SHM). These methods are generally unsuitable for resource-limited wireless sensor nodes and intensive data SHM applications. In this paper, a new data coding and transmission method is proposed that is specifically targeted at the wireless SHM systems deployed on large civil infrastructures. The proposed method includes two coding stages: 1) a source coding stage to compress the natural redundant information inherent in SHM signals and 2) a redundant coding stage to inject artificial redundancy into wireless transmission to enhance the transmission reliability. Methods with light memory and computational overheads are adopted in the coding process to meet the resource constraints of wireless sensor nodes. In particular, the lossless entropy compression method is implemented for data compression, and a simple random matrix projection is proposed for redundant transformation. After coding, a wireless sensor node transmits the same payload of coded data instead of the original sensor data to the base station. Some data loss may occur during the transmission of the coded data. However, the complete original data can be reconstructed losslessly on the base station from the incomplete coded data given that the data loss ratio is reasonably low. The proposed method is implemented into the Imote2 smart sensor platform and tested in a series of communication experiments on a cable-stayed bridge. Examples and statistics show that the proposed method is very robust against the data loss. The method is able to withstand the data loss up to 30% and still provides lossless reconstruction of the original sensor data with overwhelming probability. This result represents a significant improvement of data transmission reliability of wireless SHM systems.