This protocol details a method to perform appetitively motivated tasks in rodents to test cognitive ability. When testing cognition in animals, the simplest paradigms can potentially yield quick results with minimal investment from the experimenter. Although appetitively motivated tasks are generally learnt more slowly than aversively motivated ones, they may be essential for distinguishing the effects of a treatment on learning from its effects on aversive motivation per se. For example, if a treatment improves learning in both types of paradigm, this is better evidence that it affects cognition rather than sensorimotor processes. Rats and mice easily learn position discriminations in a T-maze, especially if multiple cues, such as different objects and floor textures in the goal arms, are provided. To start, the rodent is placed in the maze and it chooses an arm. This Trial 1, however, is the only one on which this arm will be rewarded. From now on, it must always choose the other arm. The rule is simple: for example, always turn left into the arm with diagonal black stripes on the walls and gravel glued to the floor. High levels of correct responding can be achieved within 20-40 trials. The test may therefore be particularly useful with animals of low cognitive ability, such as transgenic mice derived from some 129 or SJL strains. Once the animals are habituated, each trial should take approximately 1 min. Thus, to test ten animals for 40 trials would take around 7 h.