Due to the significant expansion of the range of problems forming the class of so-called "transport tasks", including access to non-linear ones, it is advisable to have estimates of the number of all permissible plans, among which the optimal ones are chosen.<strong >Estimates are made of a number of possible plans for solving closed transportation problems for matrices of various dimensions and structures. About fifty examples of different transport tasks were analyzed as a basis for the study. It was found that by redistributing among themselves, for example, only the capacities of manufacturers and the constancy of their total sum, the number of possible plans for the problem monotonously decreases with an increase of their relative standard deviation value. From the analysis of particular empirical dependencies for matrices of various structures, analytical generalizations are obtained for the simplest situations. Received that for tasks: (a) with dimensions (2×<em >M) and (<em >N×2) directly analytical calculation of the number of plans is possible; (b) with matrices of arbitrary sizes and structures, a computer algorithm of (<em >N×<em >M) nested cycles is required; (c) with equal powers and equal capacities, a probabilistic way of estimation is possible, the results of which correlate with the exact ones at the level of about 0.8. The task may be of interest in evaluating the effectiveness of various optimization methods.