We demonstrate the Casimir interaction between two ferromagnetic boundary surfaces using the dynamic atomic force microscope. The experimental data are found to be in excellent agreement with the predictions of the Lifshitz theory for magnetic boundary surfaces combined with the plasma model approach. It is shown that for magnetic materials the role of hypothetical patch potentials is opposite to that required for reconciliation of the data with the Drude model. PACS numbers: 78.20.Ls, 12.20.Fv, 78.67.Bf 1 The Casimir effect [1] is of much interest due to its promising multidisciplinary applications in nanotechnology, condensed matter physics, physics of elementary particles, and in gravitation and cosmology [2,3]. Many experiments on measuring the Casimir force between boundary surfaces made of different materials separated by a vacuum gap or a liquid have been performed in the last 15 years [4][5][6]. It was shown that the magnitude of the Casimir force can be controlled by using different boundary materials [7,8], phase transitions [9][10][11][12][13], and by using the boundary surfaces structured with nanoscale corrugations [14][15][16][17].An unified description of both the van der Waals and Casimir forces is given by the Lifshitz theory [18] in terms of the dielectric permittivity ε(ω) and magnetic permeability µ(ω). The role of magnetic materials in the Casimir force has been studied theoretically [19][20][21][22][23][24][25][26][27][28][29][30]. The interest stems from the possibility to obtain a repulsive Casimir force for application in micromachines. Using real magnetic materials [21,25] did not validate the early results which used constant ε and µ. As µ(iξ) can be large only at ξ < 10 5 − 10 9 Hz, its entire contribution to the Lifshitz formula is through the zero Matsubara frequency [27,28]. For metals, the zero-frequency term is strongly influenced by the inclusion (Drude model approach) or neglect (plasma model approach) of the relaxation properties of free electrons [4]. Thus using µ provides another parameter to study the role of the relaxation properties of free electrons in the Casimir effect. Some experiments demonstrate strong disagreement between the measured data and theoretical predictions when the relaxation properties of electrons are taken into account for metals [4,31,32] or the dc conductivity is included for dielectrics [4,12,13]. The same data are found to be consistent with theory when the relaxation properties are neglected for metals or the dc conductivity of dielectrics is disregarded. Two other experiments [33,34] are claimed to be in favor of the Drude model approach (see critical discussion in [35][36][37][38] Here we have used the same apparatus and cantilever preparation as in Refs. [32,40].The gradient of the Casimir force was measured acting between a Ni-coated hollow glass microsphere of R = 61.71 ± 0.09 µm radius attached to the tip of a rectangular Si cantilever and a Si plate also coated with Ni. The thicknesses of Ni coating were 210 ± 1 nm and 250 ± 1 n...