The impact of tilt-related errors on the positioning of microcantilever-based microelectromechanical systems (MEMS) on-wafer electrical probes, having multiple contact pads, is quantified and investigated here. A tilt error associated with probe roll results in the probe contact pads not being parallel to the approaching surface as a downward overtravel is imposed—this leads to one probe pad making contact with the surface before the others. In a MEMS-based probe, the analysis of the impact of roll error angle must consider both the bending and the torsion of the flexible cantilever as the overtravel is increased—something which eventually results in all pads being in contact with the surface, but not with the same contact force. An original mathematical description of the problem is presented. By making some assumptions, the analytical modelling enables the derivation of elegant equations relating the roll error angle and the cantilever deflection to achieve planarity of the cantilever apex with the underlying surface. The modelling predicts probe tip planarity for rectangular and trapezoidal shaped probes. The predictions of the modelling are tested by using macroscopic cantilevers—excellent agreement between modelling and experiment is demonstrated. The macroscopic experimental setup reveals interesting behaviour concerning a bending/twisting, tilted cantilever in contact with—and skating across—an underlying surface. The experimental findings also indicate the pertinence of the modelling for the potential use with understanding the behaviour of microscopic cantilevers—such as MEMS-based probes—similarly in contact with a surface. A flexible microcantilever enables a torsional compensation of the roll error angle. It also enables a protocol where the roll error angle can be corrected. The design geometry of the probe tip will determine which approach is best suited. In principle, the modelling is scalable to MEMS probes composed of silicon-based cantilevers.