This paper analyzes two examples that show the imprecision of the data processing by software package Statistica. The first experiment measured the intensity of soil compaction in MPa. The measurements were repeated 10 times on each of the measured positions. Position of the measurements at depths are at the distance of 15 cm. The intensity of the soil compression on the unplowed soil was measured in front of the wheels along the longitudinal axes of the tractor wheel by electronic penetrometer which has a diameter of 12.83 mm and the angle of penetration of 30°. The dependence of the intensity of soil on the compaction depth is approximated by the polynomial trend. The first example shows the inprecision in determination of coefficients in the polimomial trend. The obtained coefficient of determination gives the wrong impression about the accuracy of the obtained polynomial, or more precisely, about the accuracy of the obtained coefficients in the polynomial. After detailed revision, it was concluded that the error occurred in determining the coefficients is up to 230%. The second experiment was performed on the biomass-fired hot water boiler, located at 'Mitrosrem' agricultural company from Sremska Mitrovica, work unit Kuzmin. Nominal boiler power is 120 kW. As a fuel, there were used wheat straw bales. The impact of the quantity of inlet air (150, 220, 290, 360 and 430 m3/h) fed to the boiler firebox was continuously monitored. It was examined correlation between boiler thermal power and bale combustion time and there were given corresponding mathematical models. The software package Statistica, using nonlinear regression analysis, determined a function that approximated the experimentally obtained data. Graph of that function, drawn in the above mentioned software package, absolutely does not correspond to experimental data, although the approximate function is correctly calculated which shows a graph of the same approximations obtained by the software package Mathematica. Such errors would certainly have caused a bad interpretation of the obtained results and the impossibility of further use.