This work suggests a new algorithm for the elemental analysis, which permits taking into account both the multiplicative and additive matrix effects, a matrix composition being unknown. The use of this algorithm in the development of analysis techniques at any set of objects may appear to be successful at arbitrarily selected measurement conditions. The basis of the algorithm consists of a combination of pattern recognition and standard addition technique (PRSA). The main concept of PRSA is to predict a function approximating satisfactorily a calibration dependence for a specific object and then to determine an unknown concentration using a nonlinear addition technique, all the aforementioned being based on easily obtained information on an analyzed object and measurement conditions. The abilities of the proposed algorithm were demonstrated by the example of strontium determination by the flame atomic absorption spectrometry in superconducting systems.