Laboratory data and the results of modelling are combined to predict the possible size of craters in icy bodies such as a comet nucleus. This is done in particular for the case of a a 370-kg mass impacting a body the size of the nucleus of comet 9P/Temple-1 at 10 km s−1. This reproduces the Deep Impact comet impact to occur in 2005, when a NASA spacecraft will observe at close range an impact on the comet nucleus of an object deployed from the main spacecraft. The predicted crater size depends not only on uncertainties in extrapolation from laboratory scale and the modelling in general, but also on assumptions made about the nature of the target. In particular, allowance is made for the full range of reasonable target porosities; this can significantly affect the outcome of the Deep Impact event. The range of predicted crater sizes covers some 7–30 m crater depth and some 50–150 m crater diameter. An increasingly porous target (i.e. one with a higher percentage of void space) will increase the depth of the crater but not necessarily the diameter, leading to the possibility of an impact event where much of the crater formation is in the interior of the crater, with work going into compaction of void space and some possible lateral growth of the crater below the surface entrance. Nevertheless, for a wide range of scenarios concerning the nature of the impact, the Deep Impact event should penetrate the surface to depths of a few tens of metres, accessing the immediate subsurface regions. In parallel to this, the same extrapolation methods are used to predict the size of impactors that may have caused the features observed on the surfaces of small bodies, e.g. the Saturnian satellite Phoebe and the nucleus of comet P/Wild-2.