The amorphous state has different chemical and physical properties compared with a crystalline one. Amorphous regions in an otherwise crystalline material can affect the bioavailability and the processability. On the other hand, crystalline material can function as nuclei and decrease the stability of an amorphous system. The aim of this study was to determine amorphous content in a pharmaceutical process environment using near infrared (NIR) and Raman spectroscopic techniques together with multivariate modelling tools. Milling was used as a model system for process-induced amorphization of a crystalline starting material, α-lactose monohydrate. In addition, the crystallization of amorphous material was studied by storing amorphous material, either amorphous lactose or trehalose, at high relative humidity conditions. The results show that both of the spectroscopic techniques combined with multivariate methods could be applied for quantitation. Preprocessing, as well as the sampling area, was found to affect the performance of the models. Standard normal variate (SNV) transformation was the best preprocessing approach and increasing the sampling area was found to improve the models. The root mean square error of prediction (RMSEP) for quantitation of amorphous lactose using NIR spectroscopy was 2.7%, when a measuring setup with a larger sampling area was used. When the sampling area was smaller, the RMSEPs for lactose and trehalose were 4.3% and 4.2%, respectively. For Raman spectroscopy, the RMSEPs were 2.3% and 2.5% for lactose and trehalose, respectively. However, for the optimal performance of a multivariate model, all the physical forms present, as well as the process environment itself, have to be taken into consideration.