Quality control of metal powders in powder bed-based additive manufacturing processes is currently very time-consuming and costly and is not possible inline, i.e. during the process. The powder properties, e.g. the morphology, the rheology or the chemical composition, are of crucial importance for the quality of the components produced. For example, it is of great importance to recognize when the quality of the powder becomes too low and a powder change becomes necessary. If this happens too early, expensive powder material is lost, if it happens too late, the component quality suffers. A fast and complete inline characterization of metal powders is therefore desirable. In this poster, a method for powder quality control based on hyperspectral imaging and machine learning is presented. The aim is to qualitatively distinguish different powder types and powder batches based on hyperspectral measurements, and to quantitatively predict some powder properties. The determination of powder type and batch was possible for the investigated samples with an accuracy of up to 100 %, and good results were also achieved for the quantitative prediction of powder properties. The results show that hyperspectral imaging appears to be a promising method for inline powder characterization for additive manufacturing processes. It could therefore lead to a higher quality of the produced parts and could enable more effective and thus more ecological and economical use of the powder raw materials.