In this paper, we present a consistent analysis of the methods dealing with the derivation of recurrence relations for calculation of the T-matrix. Central and forward recurrence relations are obtained in the framework of the invariant embedding T-matrix method, the matrix Riccati equation method, and the superposition T-matrix method. The accuracies and efficiencies of the central and forward recurrence schemes are analyzed, and some implementation issues related to the improvement of the numerical accuracy and to the problem of overcoming of overflow errors are discussed.