Toxic chemicals inside building materials have long-term harmful effects on human bodies. To prevent secondary damage caused by the evaporation of latent chemicals, it is necessary to detect the chemicals inside building materials at an early stage. Deep Raman spectroscopy is a potential candidate for on-site detection because it can provide molecular information about subsurface components. However, it is very difficult to spectrally distinguish the Raman signal of the internal chemicals from the background signal of the surrounding materials and to acquire the geometric information of chemicals. In this study, we developed hyperspectral wide-depth spatially offset Raman spectroscopy coupled with a data processing algorithm to identify toxic chemicals, such as chemical warfare agent (CWA) simulants in building materials. Furthermore, the spatial distribution of the chemicals and the thickness of the building material were also measured from one-dimensional (1D) spectral variation.