Experimental evidence shows that snow is a pressure sensitive, dilatant, strain-softening material in slow, constant rate, shear deformation. When strain softening initiates in a weak layer underneath a snow slab, avalanche release is hypothesized to be possible with or without additional loading. Specifically, two cases are discussed: (1) a shear-crack-like disturbance can initiate after formation of a slip surface in the weak layer and traverse the layer by a self-propagating progressive failure with or even without loading, and (2) a self-propagating shear instability can develop when a region of the weak layer is driven past peak shear strength by loading. For these cases the implied fracture sequence, fracture geometry, time scale of release, and temperature effects are discussed and shown to be consistent with the known facts of dry slab avalanche release.