Two methods of sap flow measurement using the stem surface heat balance approach were compared: the variable power (VP) and constant power (CP) methods. Comparisons were carried out on woody stems in a glasshouse in the UK (Ficus benjamina) and in the field during the rainy season in Niger (Guiera senegalensis). Substantial errors occurred if changes in stem temperature were neglected in the CP method. The simple VP method used neglected conductive losses, but was less affected by changes in stem temperature than the CP method. Therefore, at low flows the VP method overestimated sap fluxes, but at high flows it had a better dynamic response than the CP method. Daily totals of sap flow over 9 days in G. senegalensis stems with leaf areas of 1.3-2.3 m2 ranged from 2 to 9 kg day-1, and the two methods agreed well, although there was a consistent 8-12% higher estimate by the CP method.