To find a simple means for monitoring the heat output of a wood stove, general engineering models of heat transfer are used to develop a model that predicts the heat output of a stove from measurements of surface temperature. Using the surface area and the measured surface temperature as inputs, the model predicts the heat output of the stove by radiation and natural convection. As a means of verification, surface temperature data from four wood stoves monitored in a calorimeter room are used to make heat output predictions. The predicted heat outputs are then compared with the actual heat outputs measured by the calorimeter room. The predictions involve several potential monitoring schemes:(J) separate temperature measurements for each surface of the stove,() an average temperature measurement for all stove surfaces, and () a single surface temperature measurement. The accuracies of the predictions are characterized by their geometric bias and scatter as well as their predictions of total energy delivered. The scatter is a measure of the trackability of the model, analogous to the arithmetic standard deviation. Predictions made from average temperature measurements are found to be as accurate as those based on individual temperature measurements, whereas single-temperature measurements cause an additional 5 % uncertainty in predictions. For both the average temperature and individual temperature predictions, the bias is between 2 % and 24%, with 16% as the typical scatter. The trend in the bias is underprediction, possible causes of which are discussed at length.