Some recent problems relating to work-hardening of metal crystals, such as the contribution of dislocation stress fields to the flow stress, the mechanism of the second stage, or linear, hardening, and the transition to the strongly temperature dependent third stage are discussed. It is shown that linear hardening can be explained by recourse to the strain invariance of the geometrical pattern of the dislocation network. The thermodynamic origin of this invariance is considered. Experimental data on the temperature and strain-rate dependence of the stress at which transition to the third stage occurs are shown to accord best with the view that at temperatures at which diffusion effects are negligible the third stage begins with the onset of appreciable dynamic recovery, involving the formation of point defects, probably through non-conservative movement of elementary intersection jogs.