In the forensic context, there are many challenges concerning new psychoactive. The lack of data can affect characterization and risk assessment. With the increase of new structures over the years, experimental procedures cannot provide data about them. The use of in-silico methods has proven to be an essential ally to supply information more quickly about these compounds. This work used free and paid software to obtain toxicological information and physical-chemical properties about amphetamine-like structures. We applied unsupervised (Principal Component Analysis - PCA) and supervised (Soft Independent Modeling of Class Analogy - SIMCA) to evaluate data. As a main result, we observed that both chemometric tools were able to identify different classes regarding amphetamines and cathinones. PCA showed solubility indicators as an important role in discriminating amphetamines and cathinones. SIMCA methods reproduced those clusters observed in PCA. Solubility indicators also showed the most discriminating power. However, results for classification reported similar behavior for some molecules in different classes, indicating that there are structures that can be confused regarding toxicological and physical-chemical properties. It can provide a warning about molecules, showing that they need careful attention when evaluated. These results can drive decision-making about these substances, guiding public policies regarding the prohibitions and actions aimed at medical intervention in critical cases of abuse.