The paper proposes to manage complexity and costs issues of the fault-tolerant programs not at a single program level but rather from the point of view of the whole set of such programs, which are to be run under hard time constraints. A concept of the multiple processor programs is used to model a fault-tolerant program structure. This model, in turn, is used to formulate the fault-tolerant programs scheduling problem under hard time constraints. Since the discussed problem is computationally difficult, three scheduling algorithms, based on three different metaheuristics, have been proposed. To evaluate the proposed algorithms computational experiment has been carried. The proposed global approach has been also compared with scheduling without search for the global optimum. Experiment results prove that the approach could be advantageous by producing more reliable schedules within hard time constraints.