In this paper, three techniques for determination of the sonic/catch-up points in unsteady shock reflections based on numerical flowfield analysis are considered: the Mach-number-based technique, the characteristic-based technique, and the perturbation technique. These techniques are compared using the problem of shock reflection from a convex cylinder simulated with an inviscid, non-heat-conducting flow model and an ideal reflecting surface. It is shown that the sonic points obtained with the Mach-number or characteristic-based techniques, coincide with the catch-up point obtained by the perturbation technique. The obtained sonic point converges to the theoretical sonic point given by the steady two-shock theory as the grid is refined. Quantitative data are presented, which show that very fine meshes are needed to approach the theoretical value with good accuracy. Furthermore, potential sources of significant experimental errors when applying the perturbation technique in shock-tube experiments are identified.