The Tower of London (TOL) task has been used extensively as a test of planning ability in neuropsychological patients and normal populations. Participants are asked to preplan mentally a sequence of moves to match a start set of discs to a goal, and then to execute the moves one by one. The mental preplanning stage has been identified as critical to efficient performance. The current experiments examined whether manipulations of mental preplanning influence performance on the TOL. In Experiment 1, the effect of different planning instructions was examined. Those told to make full mental plans spent considerably longer in planning than participants given no specific planning instructions, yet there was no effect of instruction condition on the efficiency of executing plans. Experiment 2 investigated whether people were able to plan mentally, by looking at their ability to identify intermediate states of an optimum mental plan. Results indicated that most individuals could make accurate preplans up to two subgoals ahead, but not three. However, making an efficient preplan did not result in better subsequent execution of moves to solve the TOL trial. It is concluded that people can make effective mental plans for a limited number of moves. However, on the TOL task, mental preplanning does not offer benefits in terms of quicker performance, or more accurate solution. The nature of planning in the TOL task is therefore questioned.