The relative merits of the design of four cubicle types, Newton Rigg, Dutch Comfort, Super Dutch Comfort and a Cantilever type were evaluated through the analysis of cattle behaviour by direct observation. The cubicles were examined in pairs over four experiments: (1) Dutch Comfort vs Super Dutch Comfort, (2) Super Dutch Comfort vs Newton Rigg, (3) Newton Rigg vs Cantilever and (4) Dutch Comfort vs Cantilever. All experiments were conducted with a one cow : one cubicle ratio. The cows were in mid-gestation, of different ages, weights and milk yields and each of the four experiments was conducted over a period of five non-consecutive days and nights.Experiment 1 compared occupancy of Dutch Comfort and Super Dutch Comfort cubicles. Experiments 2, 3 and 4 compared occupancy, preference, turnover (number of cow visits per observation period) and continuous lying time (duration of visits). The two Dutch Comfort designs had the highest occupancy and the highest turnover rates, followed by the Cantilever cubicle. The Newton Rigg model showed lowest occupancy levels and had the lowest turnover rate. Cows showed slight preferences for the Super Dutch Comfort cubicle over the Dutch Comfort model but the Dutch Comfort model was selected more frequently than the Cantilever which, in turn, was selected more often than the Newton Rigg cubicle. Once a cow choose a cubicle, the continuous lying time was always similar regardless of cubicle type.