This report describes work performed in a prestudy on how machine learning can be applied to design of radar absorbents. Radar absorbers are an important technology for reducing radar signature and thereby increasing the survivability of military platforms. Machine learning for radar-absorber design is assessed to have good potential to provide significant improvements, both by creating better absorbers and by accelerating the current design process. This research field is seen as a part of an accelerating trend where several applications within design of electromagnetic components may be improved by applying machine learning methods. The work has contributed to building expertise in applying machine learning to radar absorbers through collaboration between two departments. The focus has been on compiling the most relevant literature in the area and on building basic capability in electromagnetic simulation software. In the long term, this software can be used to generate training data for machine learning models. The report describes the common problem areas where machine learning is applied, as well as the challenges within the area. Furthermore, an approach for future work, including a methodology for applying machine learning to radar absorber design, is suggested. The target audience for this report consists of engineers and researchers interested in the design of electromagnetic structures, particularly those focused on applying machine learning to the design of radar absorbers. The conditions for designing radar absorbers using machine learning at FOI are favorable. The subject is strategically important and has the potential to deliver good results, with applications in the domains of Air Force, Armed Forces and Navy Forces.