A method for real-time scheduling of an application having a plurality m of software tasks executing at least one processing operation on a plurality N of successive data frames, each of said tasks i being defined at least, for each of said frames j, by an execution time Cij, an execution deadline Dij and a guard time Pij with respect to the next frame j+1, said guard time Pij being greater than or equal to said deadline Dij,includes: for each task i, calculating the ratios C i j D i j and then searching for the maximum over all frames j of ratios max ( C i j D i j ) , comparing the sum ∑ i = 0 m - 1 ⁢ max 0 ≤ j ≤ N - 1 ⁢ ( C i j D i j ) with the number of processors operating in parallel over which the total computation load of the real-time application is distributed, and, if said sum ∑ i = 0 m - 1 ⁢ max 0 ≤ j ≤ N - 1 ⁢ ( C i j D i j ) is less than or equal to the number of processors, concluding on the feasibility of the real-time execution of all said software tasks distributed over said processors.