According to Locard's switching theory, fiber evidence can help reconstruct criminal incidents. Generally, fiber evidence is identified by chemical composition analysis; however, the integrity of material evidence will be destroyed. For this reason, we established a testing and analysis method based on the theory of bidirectional reflectance distribution function (BRDF) to ensure the long-term retention of fiber evidence. To be specific, three groups of fiber samples were selected, with each group containing six kinds of fiber samples with the same color and different materials. The first-order differential and difference methods were used to extract spectral characteristics and analyze spectral BRDF. According to spectral curve fitting, the fitting function and residual coefficient were extracted for differential comparison. From the physical mechanism, a Davis model was built to simulate the spectral BRDF curve. As a result, the material evidence and suspicious results in the cases were effectively determined. The experimental results provide an effective and non-destructive detection technology and analysis method for the identification of fiber evidence characteristics, which greatly facilitates criminal investigation analysis and conviction.