Determination of the accuracy of commonly used techniques for measuring sap flux density in trees often presents a challenge. We therefore designed and built a verification system for heat-based sap flow sensors typically used at stem level. In the laboratory, a Mariotte's bottle device was used to maintain a constant flow rate of water through freshly cut stem segments of American beech (Fagus grandifolia Ehrh.). This verification system was used to determine the accuracy of three heat-based sap flux density techniques: heat pulse velocity, thermal dissipation and heat field deformation. All three techniques substantially underestimated sap flux density when compared against gravimetric measurements. On average the actual sap flux density was underestimated by 35% using heat pulse velocity, 46% using heat field deformation and 60% using thermal dissipation. These differences were consistent across sap flux densities ranging from 5 to 80 cm(3) cm(-2) h(-1). Field measurements supported the relative sensor performance observed in the laboratory. Applying a sensor-specific correction factor based on the laboratory test to the field data produced similar estimates of sap flux density from all three techniques. We concluded that a species-specific calibration is therefore necessary when using any of these techniques to insure that accurate estimates of sap flux density are obtained, at least until a physical basis for error correction can be proposed.