X-ray fluorescent (XRF) analysis is one of the most widely used methods for analytical control of soil and rock samples, which allows determining a wide range of elements using a single sample with minimum costs. This work is aimed to expand the XRF capabilities in terms of obtaining calibration dependences for various types of samples. The work presents the results of the experimental assessment of the accuracy of the method of fundamental parameters for analysis of soil samples on a Quant’X energy dispersive spectrometer with the UniQuant software using artificial samples. The suggested methodical approach has shown the potentiality of using a Quant’X energy dispersive spectrometer with the UniQuant software for analysis of soil samples prepared by different methods (fused, pressed and bulk). The systematic deviation from the target value and a random error were estimated. For fused soil samples, deviations from the target value for most of the major components were (1.01) with a random error of 0.01 – 0.06. A Quant’X spectrometer with the UniQuant software allows determining light elements in fused samples with an error of up to 0.06. For the majority of impurity elements, a significant deviation was found, more than one (an underestimation of the determination results by an average of 1.18 times). This requires correction of the results obtained and the analysis error will be determined by values?? ranging from 0.01 to 0.1. In most cases, the deviation from the target value and the error of the element determination decreased as the sample preparation became more complex. After adjustment, bulk soil samples showed the highest error of the analysis results (from 0.1 to 0.2 for the major elements and 0.05 – 0.10 for impurity elements). The detection limits of some elements in fused soil samples were estimated. When the component content in the soil is more than 0.1%, the best option is to analyze fused samples. To determine concentration levels of about 50 ppm, it is proposed to measure pressed samples and take into account the increased error.