In the past few years, numerical weather models (NWM) have been investigated to improve mapping functions which are used for tropospheric delay modeling in VLBI and GPS data analyses. The Vienna Mapping Functions VMF are based on direct raytracing through NWM, and so they are able to exploit the full information provided in the NWM. On the other hand, the Isobaric Mapping Functions IMF are using intermediate parameters calculated from the NWM. In this study, pressure level data from ECMWF (European Centre for Medium-Range Weather Forecasts) are applied to determine the coefficients of the VMF and the IMF. Used for the analyses of IVS-R1 and IVS-R4 VLBI sessions, both mapping functions improve the repeatability of baseline lengths (by ~10% for IVS-R1 and ~5% for IVS-R4) compared to the Niell Mapping Functions NMF.