Moisture is one of the main causes of damage to building materials (e.g. concrete/screed) and infrastructure. In Germany alone, the annual damage due to moisture exceeds several 10 billion Euro per year. However, common non-destructive measurement devices cannot reliably determine the absolute moisture content due to surface effects, cross influences and complex material dependences. Although NMR relaxation measurements are commonly used to investigate the moisture content, distribution and transport in building materials, it is usually treated as a laboratory method only. With the development of unilateral devices which allow to investigate extended samples, NMR becomes increasingly interesting for field application. For unilateral NMR the biggest challenge concerning building materials are the very fast transverse relaxation times (10μs-100ms) and a complex T2 distribution. Additionally, unilateral NMR devices usually suffer from poor SNR. This in combination with the demand to detect moisture contents down to 0.5% of dry mass, makes the fit of the decay signal and the estimation of the initial amplitude very inaccurate. Therefore, we are left to use the attenuated signals with moisture dependent relaxation times resulting in non-linear and material dependent correlation curves between the NMR signal and moisture content. In this contribution we present a method to determine such correlation curves based on a single T2 measurement close to saturation. This yields a substantial saving of time (upto several weeks) compared to a manual, empiric determination. When using a capillary bundle model and the Young-Laplace equation to describe the drying behavior of porous media, larger pores are emptied before smaller ones. Since the T2 distribution reflects the pore size distribution of the material, we introduce a moisture dependent cut-off radius r_c, and relaxation time T_(2,c), respectively. Subsequently, the expected relaxation decay at a certain moisture content can be reconstructed. Our results show a good agreement between the predicted and experimentally determined correlation curves for different building materials. The reference moisture content of the samples was determined with the Darr Method . Our method is easy to use and allows a fast build-up of a material database. Together with the technology of single-sided NMR this could be a vital step towards a reliable and consumer suitable moisture measuring tool.