Humans appear to act in response to environmental demands or to pursue self-chosen goals. In the laboratory, these situations are often investigated with forced- and free-choice tasks: in forced-choice tasks, a stimulus determines the one correct response, while in free-choice tasks the participants choose between response alternatives. We compared these two tasks regarding their susceptibility to dual-task interference when the concurrent task was always forced-choice. If, as was suggested in the literature, both tasks require different "action control systems," larger dual-task costs for free-choice tasks than for forced-choice tasks should emerge in our experiments, due to a time-costly switch between the systems. In addition, forced-choice tasks have been conceived as "prepared reflexes" for which all intentional processing is said to take place already prior to stimulus onset giving rise to automatic response initiation upon stimulus onset. We report three experiments with different implementations of the forced- vs. free-choice manipulation. In all experiments we replicated slower responses in the free- than in the forced-choice task and the typical dual-task costs. These latter costs, however, were equivalent for forced- and free-choice tasks. These results are easier to reconcile with the assumption of one unitary "action control system."