The description of the evaporation of rainfall intercepted by forests in terms of a regression of evaporation loss on incident rainfall is discussed and some of the assumptions implicit in that method are re-examined. The two major factors which control the evaporation of intercepted rainfall are identified. These are: (i) the amount of time that the canopy spends saturated during rainfall and the evaporation rate applicable under these conditions; and (ii) the canopy saturation capacity and the number of times this store is emptied, by drying out after the cessation of rainfall. A model is then constructed which is conceptually similar to the Rutter model, but which replaces that model's numerical approach with an analysis by storm events. The evaporation from a saturated canopy during rainfall is estimated from the Penman-Monteith equation; the evaporation after rain has ceased, the effect of small storms insufficient to saturate the canopy, wetting-up the canopy and evaporation from the trunks are added as separate terms. The model has been tested against data from Thetford Forest in East Anglia, with satisfactory agreement between observation and estimation. It is suggested that the model may be capable of making useful estimates of the evaporation of intercepted rainfall, solely from rainfall measurements.