Near-infrared spectrum analysis coupled with partial least squares regression can predict anaerobic digestion performances. Nonetheless, due to the complexity and diversity of organic matter, a detailed assessment of the effects of the composition of organic matter on biogas production remains a great challenge. Based on 275 samples representing a wide diversity of substrates, the application of the partial least square b coefficients to assess the effects of the involved molecules on the performances of anaerobic digestion processes is discussed. In particular, to accurately predict variables, there is a need to account for the whole near-infrared spectrum. The characterization of organic matter involving proteins, carbohydrate and lipid contents, chemical demand in oxygen, biodegradability, methane yield, and methane production kinetics data is demonstrated.