Previous research suggests that older persons show cognitive deficits in standardized laboratory tests, but not in more natural tests such as the Multiple Errands Task (MET). The absence of deficits in the latter tests has been attributed to the compensation of deficits by strategies based on life-long experience. To scrutinize this view, we primed older participants with positive or negative stereotypes about old age before administering MET. We found that compared to unprimed controls, priming with positive age stereotypes reduced the number of errors without changing response times, while priming with negative stereotypes changed neither errors not response times. We interpret our findings as evidence that positive age priming improved participants' cognitive functions while leaving intact their experience-based compensation, and that negative age priming degraded participants' cognitive functions which, however, was balanced by an even stronger experience-based compensation.