The dynamics of magnetic field lines and of charged particles in toroidal chambers are commonly analysed by numerically solving the dynamical equations. They may also be analysed using deterministic reduced models, i.e. low-dimensional discrete time approximations (maps) of the Hamiltonian continuous time models. We report on the accuracy of the latter method by considering the mapping technique derived from the Hamilton–Jacobi equation. The optimum time stepping in some models for the study of the magnetic field in tokamaks is determined by using local criteria. Special attention is given to the analysis of stochasticity produced by time discretization.