The most common housing system for reproduction rabbits, individual cage housing on a wire floor, is increasingly scrutinized because of its potential detrimental impact on animal welfare. We compared three types of housing: (1) individual cage housing on a wire floor (3952 cm2/doe, maximum roof height 63 cm, one 1000 cm2 plastic footrest/doe), (2) semi-group housing on a wire floor (5000 cm2/doe, roofless, one 1000 cm2 plastic footrest/doe) and (3) the same semi-group housing, but with a fully plastic slatted floor. In all housing systems, does had free access to an elevated platform. In the semi-group housing pens, four does were housed communally during 21 days of the reproduction cycle (to allow more space for locomotion and to increase opportunities for social contact), and individually during the other 21 days of the cycle (to minimize doe–doe and doe–kit aggression that peaks around kindling). In all, 24 Hycole does were included per system. The does entered the experiment at 203 days of age (after their first parity). The experiment consisted of four reproductive cycles, ending at 369 days of age. Pododermatitis was scored in cycles 1, 2 and 4. At the end of the 4th cycle the does were euthanized and X-rays were taken to assess spinal deformation. Tibia and femur length, width and cortical thickness were determined and bone strength was assessed using a shear test, as a measure of bone quality. Although severe pododermatitis was absent, the prevalence of plantar hyperkeratosis (hair loss and callus formation) at the end of the 4th cycle was much greater on the wire floor (65% and 68% for semi-group housing and individual cages, respectively) than on the plastic floor (5%, P<0.0001), even though the wire floors were equipped with a plastic footrest known to decrease hyperkeratosis. In contrast to our expectations, semi-group housing did not affect the prevalence of spinal deformations (P>0.10), but in line with our expectations bone quality was affected favourably by semi-group housing. The tibial cortex (and to a lesser extent the femoral cortex) was thicker in semi-group housing than in individual cages (1.45, 1.46 and 1.38 mm for semi-group housing on wire, semi-group housing on plastic and individual housing on wire, respectively, P=0.045). What this increase in cortical thickness means in terms of doe welfare requires further study, as it may reflect an increase in activity resulting either from increased space for locomotion, or from fleeing aggressive pen mates.