Evaporation from wet canopies (. E) can return up to half of incident rainfall back into the atmosphere and is a major cause of the difference in water use between forests and short vegetation. Canopy water budget measurements often suggest values of E during rainfall that are several times greater than those predicted from Penman-Monteith theory. Our literature review identified potential issues with both estimation approaches, producing several hypotheses that were tested using micrometeorological observations from 128 FLUXNET sites world-wide. The analysis shows that FLUXNET eddy-covariance measurements tend to provide unreliable measurements of E during rainfall. However, the other micrometeorological FLUXNET observations do provide clues as to why conventional Penman-Monteith applications underestimate E. Aerodynamic exchange rather than radiation often drives E during rainfall, and hence errors in air humidity measurement and aerodynamic conductance calculation have considerable impact. Furthermore, evaporative cooling promotes a downwards heat flux from the air aloft as well as from the biomass and soil; energy sources that are not always considered. Accounting for these factors leads to E estimates and modelled interception losses that are considerably higher. On the other hand, canopy water budget measurements can lead to overestimates of E due to spatial sampling errors in throughfall and stemflow, underestimation of canopy rainfall storage capacity, and incorrect calculation of rainfall duration. There are remaining questions relating to horizontal advection from nearby dry areas, infrequent large-scale turbulence under stable atmospheric conditions, and the possible mechanical removal of splash droplets by such eddies. These questions have implications for catchment hydrology, rainfall recycling, land surface modelling, and the interpretation of eddy-covariance measurements.