We present an associative recognition experiment comparing three samples of healthy people (young people, older people with high cognitive reserve [HCR], and older people with low cognitive reserve [LCR], with each sample consisting of 40 people), manipulating stimuli repetition during the study phase. The results show significant differences among the three samples in their overall performance. However, these differences are not due to a different use of familiarity, but rather due to a different way of using recollection: although there are no differences in the hit rates between the HRC and LRC samples, the LCR group makes significantly more recollective false alarms than the HCR group. Moreover, repetition provokes an increase in the recollective false alarms in the LCR group, but this does not occur in the group of young people or in the HCR group. These findings are explained in terms of recollection-based monitoring errors and seem to provide support for the cognitive reserve hypothesis.