An objective need for a discussion on the implementation of the method of experimental study of the magnetic properties of ferromagnetic materials with the use of “long” samples from these materials in the appropriate diagnostics (alternative to the method using classical toroidal samples) and argumentation of their sufficient relative length is shown. The question is raised about the consistency of the existing concept (postulated) concerning the required value of the ratio of the length of a rod sample of a cylindrical shape to its diameter L/D = [L/D] = 50 (starting from which, the magnetic parameters of the sample and its material should correspond to each other). To substantiate the choice of the criterion value L/D = [L/D], families (by L/D) of field dependences of magnetic induction, permeability, susceptibility, and magnetization for cylindrical steel samples were obtained, and on their basis and being very informative for this purpose, families of dependences of such parameters from L/D in the range L/D = 1–50 are determined. This made it possible to identify the values of L/D = [L/D] by the characteristic output of these dependences in the self-similar region. At the same time, a previously unreported fact was established that the values of L/D = [L/D] depend on the intensity of the magnetizing field H (and in a more general case on the magnetic properties of the material, in particular its magnetic permeability μ); hence, the criterion values [L/D] are as follows: from [L/D] = 10–15 at H = 54.2 kA/m (μ ≅ 31) to [L/D] = 50–60 at H = 4.7 kA/m (μ ≅ 272) and at H < 4.7 kA/m (μ > 272), [L/D] > 50–60, which is higher than the commonly believed value. Phenomenological expressions are obtained for [L/D], with arguments of both H and μ. It is also shown that for the practically used L/D range of samples, the exponential relationship between their demagnetizing factor and the power function of the L/D parameter is valid.