The inverted-pendulum method of thrust measurement for high-power electric thrusters is described analytically and experimentally. Two sources of uncertainty in the thrust measurements are investigated: those due to changes in the sensitivity of the inverted pendulum and those due to tare forces resulting from thermal stresses and electromagnetic forces acting on the flexural elements. An idealized analytical model shows that changes in sensitivity occur due to changes in the mass loading of the inverted pendulum and the temperature of the flexure elements. The sensitivity can be increased only at the expense of more stringent requirements on mass loading and thermal control. The results of the model are validated with measurements made using a carefully calibrated inverted pendulum thrust stand having several advantageous design features including gallium pots for current conduction, an optical two-axis displacement sensor, and careful consideration of flexure shape and cooling requirements. The sensitivity is shown to be constant over a wide range of operating conditions. Measurements show that thermal drifts of the stand are linear on the time scale of the thrust measurements and are less than 0.6% of the expected thrust. The thrust stand operation was demonstrated during experiments with a 30-kW lithium-fed applied-field magnetoplasmadynamic thruster. Under nominal conditions (400 A, 0.1 T, 17.5 mg/s) the thermal and electromagnetic tare forces were measured at less than 9% of the thrust and the total thrust uncertainty at ≤ 10% of the thrust.