The original stochastic interception model (Calder, 1986, J. Hydrol., 89: 65–71) showed, through the use of Poisson probability statistics, that the rate at which vegetation canopies are wetted is dependent on the volume of the individual raindrops; for the same depth of rain applied, a greater depth of water will be retained on the canopy when raindrop volumes are small. This paper recognizes that vegetation canopies are wetted through both the primary impact of raindrops to the top layer of the canopy and secondary impacts, from drops falling from the vegetation, to lower layers of the canopy. It shows how drop volumes of primary raindrops can be calculated from the Marshall-Palmer distribution and drop volumes of secondary drops can be calculated from the `characteristic' volume appropriate to the particular vegetation species. A two-layer stochastic model is described which calculates the rate at which canopies are wetted in relation to both primary and secondary drops. The paper also shows that, in addition to the volume-dependent stochastic wetting effect, there is also another drop-size-dependent wetting effect, related to the kinetic energy of the raindrops, which influences the maximum storage that can be achieved on the canopy. The wetting functions predicted for canopies of different density are described and the application of the model for assessing interception loss from forests, in relation to vegetation, climate and possible climatic change, is discussed.