Nitrocarburization is a process for surface hardening of metallic components. For quality control purposes, regular inspection of the achieved hard surface layer, known as the compound layer thickness, is necessary. This is done by cutting the samples and inspecting a cross-section with microscopy. Since this is a destructive and time-consuming technique, it limits the frequency of the testing, and hence there is a need to complement this method with a rapid, non-destructive alternative. This paper demonstrates how ultrasound in combination with a supervised learning approach, can be used to estimate compound layer thickness down to 10-20 microns.