The apparent shear stress from plate-plate magnetorheometry, using the commercial magnetocell MRD180/1T (Physica/Anton Paar) in standard configuration, is distinctly overestimated. The effect is due to a flux density maximum near the sample rim and radial migration of iron particles towards the rim. Radial magnetic flux density profiles were investigated by finite element simulations using the Maxwell (R) 2D code and by direct Hall probe measurements. The reliability of the finite element method results, both for the empty magnetocell and with magnetorheological fluid (MRF) in the measuring gap, allows conclusions on the true flux density within the MRF, which cannot be accessed by Hall probe measurements. If the MRF sits on top of the bottom yoke (standard configuration), the flux density maximum reaches twice the plateau value (0.74 T for 3 A coil current and 0.3 mm gap height of the investigated MRF). This yields a higher effective flux density and causes radial iron particle migration, resulting in an additional magnetic flux increase near the rim due to augmented MRF magnetisation. As a result, the rotor torque at constant rotary speed increases with time. Reliable results are achieved by a modification of the magnetocell, such that the MRF sits on a non-magnetic Hall disc of 1.5 mm thickness, allowing an online flux density measurement by a FW Bell Hall probe. In this configuration, the radial flux density profile near the rim remains sufficiently flat and no iron particle migration is detected.