The performance of a skin evaporimeter is analysed in terms of three resistances to the diffusion of water vapour. The instrument underestimates evaporation when the resistance of the probe r_p is greater than the boundary layer resistance it replaces r_a. The underestimation is trivial and the manufacturer's calibration is valid when the skin resistanee r_s is much larger than r_p and r_a—a condition satisfied when evaporation is close to the basal rate. Very rapid evaporation, e.g. from burns, is underestimated because the presence of the probe restricts the flux of vapour when r_s is of the same order as r_p. An empirical correction is given for arbitrary values of r_a.