MEMS thermal shear-stress sensors exploit heat-transfer effects to measure the shear stress exerted by an air flow on its solid boundary, and have promising applications in aerodynamic control. Classical theory for conventional, macroscale thermal shear-stress sensors states that the rate of heat removed by the flow from the sensor is proportional to the 1/3-power of the shear stress. However, we have observed that this theory is inconsistent with experimental data from MEMS sensors. This paper seeks to develop an understanding of MEMS thermal shear-stress sensors through a study including both experimental and theoretical investigations. We first obtain experimental data that confirm the inadequacy of the classical theory by wind-tunnel testing of prototype MEMS shear-stress sensors with different dimensions and materials. A theoretical analysis is performed to identify that this inadequacy is due to the lack of a thin thermal boundary layer in the fluid flow at the sensor surface, and then a two-dimensional MEMS shear-stress sensor theory is presented. This theory incorporates important heat-transfer effects that are ignored by the classical theory, and consistently explains the experimental data obtained from prototype MEMS sensors. Moreover, the prototype MEMS sensors are studied with three-dimensional simulations, yielding results that quantitatively agree with experimental data. This work demonstrates that classical assumptions made for conventional thermal devices should be carefully examined for miniature MEMS devices.