Context. When designing various measuring instruments, the problem arises of constructing the so-called calibration characteristic of a measuring device, i.e. the quantitative dependence of the measurement result on the measured value. This characteristic is inverse to the direct characteristic – the dependence of the measured value on the measurement result. This problem is solved on the basis of approximate data obtained during the experiment with the measuring instrument. A new method for solving this problem is proposed, based on the apparatus of interval mathematics. Objective. The aim of the work is to develop a completely formalized method for constructing the calibration characteristic of a measuring instrument from approximate data obtained in the experiment with this instrument. Method. The method proposed in this article consists in presenting the function of direct con-version of a measuring device in the form of a linear interval function, determining its interval parameters (coefficients) from experimental data and solving the resulting interval dependence between the measurement result and the measured quantity with respect to the measured value. The method of solving interval equations is used. Result. General formulas are obtained that determine interval calibration characteristic of the measuring instrument on the basis of data obtained in the experiment with the instrument. A detailed analysis of formulas is performed. General laws are established that obey direct and inverse (calibration) characteristics of measuring instrument, as well as the relationship between direct and inverse characteristics (if the instrument is linear transformer). Conclusions. The article proposes a new approach to the construction of calibration characteristics of measuring instruments, based on use of interval mathematics, for processing data from experiments with instruments. This approach, unlike existing ones, makes it possible to build calibration characteristics of measuring devices and analyze them purely analytically.