The paper explains the application of regression analysis for prediction of processes in ecological systems and control of them. Theoretical analysis of the regression method structure and the main typical dependences between dependent and independent variables used for mathematical prediction modelling for statistically random processes are given. Specific aspects in significance of regression equation parameters are defined in statistical processing of the experimental data. Equation demonstrating the soil transpiration as a function of changes in fluid pressure and relative water vapor saturation is given. Comparative analysis of the absolute and relative errors of the actual and predicted values of the soil transpiration is presented. Expressions allowing one to estimate the arithmetic mean and the mean square deviation of the soil transpiration depending on the computational points are demonstrated. A method for recognising miscalculations when processing experimental data using the STATISTICA program is presented. Using the 3σ method and the tabular method, a numerical example of whether any result is miscalculated is demonstrated. Efficiency of the proposed mathematical model is shown using a specific example with account of statistical analysis of experimental results for metrological validation of them.