This paper considers two sources of error in thermogravimetry which previously have received insufficient attention, namely, instability in the inclination of microbalance supporting structures and specimen temperature measurement. For the Cahn RH microbalance, for example, which is basically a beam system dynamically balanced by an electromagnetic torque, principles of mechanics are applied to a simple model to obtain the functional variation of the microbalance output in terms of weight loads, geometric parameters and the inclination of the balance supporting structure. Together with this treatment, measurement of apparent weight changes as a function of microbalance inclination allows limits to be established for the weight errors due to changes in inclination. For the Cahn RH balance, this limit can be as large as 600 mu g mrad-1. In relation to specimen temperature measurement, uncertainties in thermogravimetrically-derived rate constants for alloy oxidation resulting from uncertainties in temperature are considered, and techniques to minimize these uncertainties are discussed.