An axisymmetric finite element model of heat flow was used to evaluate some of the assumptions inherent in the steady-state heat balance method for measuring sap flow in herbaceous plants. Results indicate that the gauge slightly overestimates conduction up and down the stem when sap flow is nearly zero, causing a corresponding underestimate of the sheath conductance and the radial outward heat flux. As sap flow rates increase, the temperature distribution in the stem and gauge is altered to the point that the one-dimensional Fourier equations are no longer applicable and the individual heat fluxes in the system are poorly estimated. However, the errors are largely self-compensating, so that the resulting gauge estimate of the heat absorbed by the sap stream is reasonably accurate. The model indicates that stem vascular anatomy affects the accuracy of the method, predicting that, in general, the method should be more accurate with dicots than with monocots.