Management strategies like “deficit irrigation” in wine grapes require accurate, reliable information on vine water use, making direct measurements of vine transpiration highly desirable. The heat-balance sap flow method has the advantages of being non-invasive and requiring no other calibration beyond a zero-flow set. Potential violations of the method's assumptions were dealt with and the heat balance method successfully applied to mature grape vines under conditions of extremely high sap flow. Greenhouse studies suggested that vines transpire at night, up to 9.5% of the total 24-h water loss, thus violating the zero-flow assumption for setting the gauge constant. Using a predetermined gauge constant caused smaller errors than using daily, pre-dawn constants set in situ. The steady-state assumption was violated only in early and late hours of the day, and the inclusion of a term to account for the change in heat stored by the stem only marginally improved daily estimates of water use. The assumption of radially uniform temperature across the heated stem segment is violated at very high flows (e.g., >700 g·h–1), but can be corrected for by using wider heaters and adjusting the placement of thermocouples. For a mature, potted vine in the greenhouse, the maximum absolute error in cumulative daytime water use between a sap gauge and a precision load cell was about –10%, with the gauge almost exclusively underestimating water loss. A custom-built, 20-gauge system was run continuously in the field for 90 days. Vine-to-vine variability in water use was not accounted for by normalizing sap flow by leaf area, suggesting that it is critical to include in any field study the largest number of gauges that are technically feasible.