Non-metallic inclusions strongly affect the quality and the properties of steels. The final inclusion landscape is essentially influenced by the secondary metallurgical treatment. By means of Optical Emission Spectrometry with Pulse Discrimination Analysis (OES-PDA), it is possible to vaporize and analyze a small volume of the steel sample including matrix and inclusions. The chemical compositions of the single sparks give an insight into the number, the size and the chemical composition of the inclusions. The short time effort for the analysis in combination with the automated application enable the generation of an extensive amount of data, which provides an important basis for the characterization of steel cleanness. The first part of this master thesis deals with the fundamentals of OES-PDA-analysis. Additionally the possibilities of data processing, the usability and validity of different specific assessment values are discussed in detail. On this basis, examinations of OES-PDA-data collected from tundish samples at Hüttenwerke Krupp Mannesmann GmbH are done. One aim of the study is the characterization of the inclusion landscape in three groups of steels with different desulphurization behaviour. Comparing the groups, the results show notable differences for the investigated groups and reflect the significant influence of secondary metallurgical treatment on the final steel cleanness. Especially the addition of fluorspar to the desulphurization slag leads to a decreasing number of Al-containing inclusions, but increases the size of the remaining particles. Further examinations deal with the connection between the removal of deoxidation products and the porous plugs availability. Finally, the modification of non-metallic inclusions with increasing dwell time between vacuum treatment and continuous casting is investigated.