One of the main tasks of geodesy is the definition and realization of reference systems. A global terrestrial reference system is realized by a reference frame of a set of positions which respect to a reference epoch and linear motions of a network of globally distributed stations on the Earth's surface. Different space geodetic techniques can be combined to realize a reference system with optimal accuracy, stability and consistency. As such, the focus of this thesis was to apply GPS and GLONASS of the Global Navigation Satellite System (GNSS) and satellite laser ranging (SLR) to LAGEOS-1 and LAGEOS-2 to determine an improved global terrestrial reference frame. The data are daily for GNSS and weekly for SLR normal equation systems over a time span of 17 years (1994 - 2010), produced from a homogeneous reprocessing. This was done through a joint effort of TU Munich, University of Bern, ETH Zurich and TU Dresden using common state of the art reducing models according to the IERS conventions. The same processing software was used to evaluate and combine the different geodetic systems to ensure the highest consistency. Utilizing the recommendations of the IERS conventions 2010, the displacement of the Earth's surface due to mass variations in the atmosphere and in the ocean was reduced from the observations. A model of the non-tidal part which is also used for the reduction of observations of the geodetic gravity mission GRACE (Gravity Recovery and Climate Experiment) was applied. The reduction of this model enhances the position accuracy of the GNSS and SLR position time series. Because of the systematic effect on SLR observations (blue sky effect) the surface load deformation should be reduced especially when combining SLR and GNSS to realize a global terrestrial system. In addition there are more geophysical effects on the station positions, for example the deformation of the Earth's surface due to continental hydrological loading. If such effects are not considered in the estimation process of geodetic observations, residual deformations are present in the estimated parameters. Therefore, this effort included modeling of deformation in the solution by using a consistent spherical harmonic approach of degree-one surface load coefficients. The residual deformations such as hydrological loading modeled with degree-one surface load coefficients have a strong annual signal for which the GNSS-only and the SLR-only solutions show the same variations. The combination of GNSS and SLR was done at the level of normal equations. Considering the strengths of each technique, an optimal weighting based on more realistic uncertainties was applied to the data. The pole coordinates and the degree-one surface load coefficients were combined. Unique to the most recent realizations of a global terrestrial reference system, no local ties (LT) of co-located sites were used in the combination of the different techniques. Thus, using a global solution together with an appropriate definition of the geodetic datum of the combined station network it was possible to estimate components of LT. This estimation enables an independent validation of the measured LT which are a crucial point in combination of different geodetic techniques.