The paper gives a brief review of mathematical models of potential composition of oil fractions. The stages of obtaining approximating dependence of the fractional composition of oil are defined. A multicomponent mixture is represented as a discrete series of narrow hydrocarbon fractions. Each of them is characterized by its average boiling point. The results of approximation obtained for oil of different fields are given. The criterion for choosing dependence is the sum of squared deviations of factored values from experimental (reference) values. The following approximating dependencies of the curve of true boiling points for oil of different deposits are obtained. They are cubic parabola, power function, arctangence, exponentiation function, cubic parabola-arctangence. It is shown that the best results are obtained for combined ‘cubic parabolaarctangent’ model. In case of data deviation on oil fractional composition coming from chemical analytical laboratory coefficients are corrected in TBP approximation. Oil fractional composition awareness is important when choosing the direction of oil refining, in process and simulation modeling, when determining the top performance of processing units.