A crystal plasticity model for hcp materials is presented which is based on dislocation glide and pinning. Slip is assumed to occur on basal and prismatic systems, and dislocation pinning through the generation of geometrically necessary dislocations (GNDs). Elastic anisotropy and, through the coupling of GNDs with slip rate, physically-based lengthscale effects are included. A model polycrystal representative of the alloy Ti–6Al, which shows creep and strain rate effects at 20 °C, is developed and it is shown that the primary effect of elastic anisotropy during subsequent plastic flow is to increase local, grain-level, accumulated slip. Lengthscale effects, however, are shown to lead to quite considerable increases in grain-boundary stresses, and to the re-distribution of accumulated slip local to grain boundaries. In particular, an initially highly non-uniform slip distribution tends to be made more uniform through the hardening effect of sessile GNDs at grain boundaries. The concept of a rogue grain combination is presented; that is, a ‘primary’ grain with c-axis orientation at or near parallel to macro-level loading, together with adjacent grains with a prismatic plane orientated at about 70° to the normal to the load direction. This particular combination of orientations leads to the highest stresses normal to the primary basal, together with high levels of accumulated slip in adjacent grains. The presence of a rogue grain combination in cycles both with and without cold dwell is examined. The cycle with cold dwell is shown to be the more damaging, and a possible criterion for facet nucleation in Ti-alloys is introduced.