A frequent assumption in the area of prospective memory is that a reminder to do an activity in the future improves the likelihood of doing the activity. The results of four experiments indicated, however, that the most general version of this assumption is incorrect. Subjects were either reminded of a prospective memory task several times during a retention interval or not reminded of the prospective memory task. The most effective reminders referred both to the prospective memory target events and to the intended activity. Reminders that referred only to the target events did not improve prospective memory (relative to a no-reminder control). Reminders that referred only to the intended activity did improve prospective memory, but not to the level of reminders that referred both to the target events and to the intended activity. Instructions to imagine oneself performing the prospective memory task did not further improve prospective memory. Neither the delay between the prospective memory instructions and the prospective memory cover task nor the delay between a reminder and a prospective memory target event significantly influenced performance. The results, which are discussed in terms of theoretical and practical implications, support a new theory of prospective memory and suggest surprising conditions under which reminders fail to benefit prospective memory.