Abstract: A preference test was used to demonstrate that gilts have the ability to associate two sets of neutral cues with two different periods of confinement and water deprivation and to anticipate the long-term consequences of their choice in the test. Twelve gilts housed in two large, straw-bedded pens were trained to go to two sets of 12 crates, positioned on each side of a choice point, for feeding twice a day. Following initial training, the two sets of crates were marked with contrasting visual patterns and the patterns were associated with either 30 min (`short' confinement) or 240 min (`long' confinement) of confinement in the crates after entry. During 16 days of preference testing, the gilts were sent alternately to one side or the other in the mornings and allowed to choose in the afternoons. Eight gilts chose the short confinement side more often, two, the long confinement side more often and two, each side an equal number of times, indicating that most gilts learned the association and preferred to be released shortly after feeding. However, gilts still chose the long confinement side on occasion, suggesting that they did not find 240 min of confinement very aversive. When the gilts were sent to the crates in the morning, their behaviour indicated that they expected to be released or confined depending on which crate they were in. The cognitive abilities of animals with respect to perception of time and anticipation of future events have important implications for their welfare. This study demonstrates that methods can be developed to ask animals about such things.