The rare earths doped glasses have been considered as the best solution to fulfil the increasing demand for cheaper fluorescent taggants. The ensuring a high quality of a spectral taggant batch imposes, among others, an exact chemical composition measurement. The most convenient method for elemental analysis of powdered specimens is Energy Dispersive (Polarised)- X-ray Fluorescence Spectroscopy (ED(P)-XRFS) because it has many advantages: non-destructive, large analytical range, robustness, easy sample preparation, cheaper consumables, less time consuming etc. ED(P)-XRFS has some inherent analytical drawbacks such as: interelements interferences, matrix dependence etc. The only way to overcome the drawbacks is a proper method for the measurement uncertainty (MU) estimation of the outcomes. Consequently, the paper argues that the bottom-up method for MU estimation is feeble as long as the sensitivity coefficients of the factors of uncertainty budget are not well known and documented, which is the habitual case of the ED and WD XRFS bottom-up published methods. The paper substantiates a new robust top-down method for MU estimation of the ED(P)-XRFS outcomes carried on a fluorescence glass. Also, the paper presents a representative case study of appliance of the developed top-down method. The other novelties addressed in the paper are: i. Introducing the robust statistics as further checks for the accuracy of mean and of standard deviation, according to ISO 13583:2015; ii. A new criterion for testing the normal distribution of the mean of a set of fewer outcomes based on Central Limit Theorem and on the central momentum of third orde; iii. The method was implemented for multiple simultaneous outcomes. The top-down method, given in the paper, can be applied to ED- and WD-XRFS routine measurements in industrial laboratories or in material science ones, but it can be adapted, without much effort, to other types of tests e.g. hardness, XRD etc.