Cognitive flexibility in non-human primates is traditionally measured with the conceptual set shifting task (CSST). In our laboratory, Guinea baboons (N=24) were continuously tested with a CSST task during approximately 10 years. Our task involved the presentation of three stimuli on a touch screen all made from 3 possible colours and 3 shapes. The subjects had to touch the stimulus containing the stimulus dimension (e.g., green) that was constantly rewarded until the stimulus dimension changed. Analysis of perseveration responses, scores and response times collected during the last two years of testing (approximately 1.6 million trials) indicate (1) that the baboons have developed an “expert” form of cognitive flexibility and (2) that their performance was age-dependent, it was at a developing stage in juveniles, optimal in adults, declining in middle-aged, and strongly impaired in the oldest age group. A direct comparison with the data collected by Bonté, Flemming & Fagot (2011) on some of the same baboons and same task as in the current study indicates that (3) the performance of all age groups has improved after 10 years of training, even for the now old individuals. All these data validate the use of non-human primates as models of human cognitive flexibility and suggest that cognitive flexibility in humans has a long evolutionary history.