Input parameters of mathematical models applied for accident reconstruction are seldom known exactly. It means that result of calculation also is given with some uncertainty, which can be estimated using total differential method. Usually applied formula contains only elements of first order Taylor expansion. However if higher order partial derivatives with respect to parameters exist and uncertainties of input parameters are large, then one can expect, that neglected remainder of Taylor expansion will have significant value. Analysis of this problem is presented in the paper. It is shown that use of total differential formulae with second order elements of Taylor series can be advisable in
the case when uncertainties of parameters changes exceeding 10%.