Earlier studies have shown that cattle preferentially utilised Dutch Comfort (DC) cubicles more frequently than Newton Rigg (NR) cubicles. Designs differ in position of barwork with the Dutch Comfort providing greater lateral space sharing than the older Newton Rigg cubicles. To further investigate the role of cubicle design on cattle behaviour and to attempt to improve utilisation of the Newton Rigg design, manipulation of cubicle frames and bedding was undertaken over a 5 month winter housing period. Forty cows in mid-gestation of differing ages, body weights and milk yields were monitored using standard behavioural observation methods. Newton Rigg cubicles were manipulated by altering the presence and location of head and breast rails and cattle utilisation was compared with the standard Dutch Comfort cubicles. This manipulation affected longitudinal space in the cubicle. Experiments were also conducted to examine the influence of mats on the occupancy levels of cubicles.In the frame manipulation experiment, occupancy rates were higher in the DC than in the NR cubicles. Removal of both head and breast rails tended to increase occupancy levels of the NR cubicles. Overall, 54% of the cows spent more time (P < 0.01) lying in the DC cubicles, whereas only 23% of the animals spent more time (P < 0.01) lying in the NR cubicles. Other information extracted from the data indicated a greater turnover (number of cows occupying individual cubicles) of DC cubicles per night than NR cubicles (P < 0.01). The mean continuous lying time (+/- SD) per cubicle per night was 106.20 +/- 93.45 min with a range of 15-540 min, regardless of cubicle type or manipulation. Cubicle fidelity was low despite evidence of preference for stall type. The mean number of cubicles visited per cow was 19.7 over the experimental period (range 13-31 ). The cows still chose DC cubicles more frequently than NR even when relative availability of NR: DC was as high as 2 : 1. The addition of mats on cubicle beds increased the occupancy of both DC and NR over standard cubicles with concrete floors, but overall, the DC cubicles retained a higher level of occupancy (P < 0.001).