A theoretical and experimental analysis of the systematic errors in the measurement of leaf temperature with thermocouples has been made. These errors are due to radiative heat exchange with the environment and to convective heat exchange with the ambient air. The magnitude of the error depends on the relevant temperature differences and on the resistance to heat transport between the object and the junction. The irreproducibility of this resistance causes the large variability in the measured temperatures. Measurement errors up to 7°C or to 30 % of the temperature difference between object and environment are found with thermocouples of 0.1 mm diameter. The radiation and convection error can be minimized by a reduction of the wire diameter and a careful positioning of the junction against the leaf surface with a slight pressure permanently exerted on it.