Steel manufacturers have made great efforts to produce as clean as possible steel. However, classic methods for the characterisation of steel cleanness are time consuming and cannot be adapted to fast process control. Moreover, while sufficient results can be obtained for macro-cleanness, their spatial resolution is poor and does not fir for size determination, chemical composition and non-metallic inclusion frequency. First attempts with a commercial laser spark showed encouraging results and, on laboratory scales, laser sources have demonstrated that the composition of inclusions can be very well characterised. Aim of the project is to develop a fast and robust laser- method characterising inclusions over a large area or volume that can be integrated into the process control.