Hydrological and land surface models require simple but accurate methods to predict the solar radiation transmitted through vegetation to snow, backed up by direct comparisons to data. Twenty shortwave pyranometers were deployed in forest plots of varying canopy structures and densities in sparse birch forest near Abisko, Sweden, in spring 2011 and mixed conifer forest near Sodankylä, Finland, in spring 2012. Above-canopy global and diffuse shortwave irradiances were also measured. These data were used to test a model that uses hemispherical photographs to explicitly estimate both diffuse radiation and direct beam transmission, as well as two models that apply bulk canopy parameters and versions of Beers Law. All three models predict canopy shortwave transmission similarly well for leafless birch forest, but for conifers, the bulk methods perform poorly. In addition, an existing model of multiple reflections between canopy and snow was found to be suitable for birch, but not conifers. A new bulk approach based on empirical relationships with hemisphere-averaged sky view fraction showed improved performance for both sites; this suggests benefits of avoiding the use of plant area index calculated from optical methods, which can introduce errors. Furthermore, tests using common empirical diffuse radiation models were shown to underestimate shortwave transmission by up to 7% relative to using the data, suggesting that new diffuse models are required for high latitudes. Copyright © 2013 John Wiley & Sons, Ltd.