Neodynium doped phosphate glass laser has been chosen as the gain medium of large laser facilities devoted to ignition such as NIF or LMJ  because it can be cast into a great variety of form and size with excellent homogeneity at relatively low cost. In this respect Megajoules class laser design relies on previous experimental and numerical studies on energy storage and gain capabilities. These parameters are particularly needed as inputs in the modeling of the amplifying stages in order to dimension and caracterize laser systems. Although there is a abondant litterature on Nd:phosphate glass, we found it necessary to make absolute value measurements. We have so developed different experimental scheme based on amplifiers pumped by a laser [2, 3] and by flashlamps . In parallel, we have developed numerical models which take into account the exact configuration of the amplifiers. The main result of these different experiments is that the use of an adjustable parameter (so-called arbitrary quantum yield) is always needed to obtain a good agreement between experimental and numerical results on gain measurements. Whatever method of pumping was used, we had to introduce a quantum yield close to 0.8 [2, 3, 4] which encompasses all unknown loss mechanisms and all hypothesis assumed in the models (section 2). We have analysed the different energy pathways and we put into evidence loss mechanisms correlated to spectroscopic properties of neodymium ions in host glasses. These mechanisms are energy transfer up-conversion and Excited State Absorption (ESA) of both pump and laser radiation (amplified pulse or florescence from the upper laser state) [6, 7]. The two latter processes are correlated to a high population of excitated states, but all of them can significantly affect the energy storage efficiency and the gain capability of these laser systems by depopulating the upper laser level or reducing the state lifetime. These mechanisms have already been reported for some materials such as neodymium doped fluorite host medium  or crystals [7, 9]. Surprinsingly enough few works have been published in case of neodymium doped glasses . Our goal in this work, was to identify and quantify each spectroscopic process that could exist in Nd :doped glasses. We report here spectroscopic experiments and the use of absolute characteristic values as new inputs in the modeling of gain capabilities of a Nd :phosphate amplifying stage (section 3). However part of the losses is also due to non spectroscopic factors such as the pumping geometry. For instance, in case of slab transversally-pumped, not taking into account Fresnel reflectivity could be at the origin of strong losses. This is the reason why we have developed in collaboration with the Lawrence Livermore National Laboratory (LLNL), a full 3D ray trace code that we have used to simulate gain measurement (section 4) . Note that this study is of larger interest for NIF-LMJ program since rare earth doped-glasses appear also promising for future broad-band solid-state systems which would be diode-pumped.