At present, a significant weakening of the intensity of transverse mixing at the confluence of large rivers, which is observed in a number of cases, is widely discussed. Since the observed features of the confluence of large watercourses are not only of research interest but also of significant economic importance associated with the characteristics of water management at these water bodies, a large number of works are devoted to their study. Water resources management requires measures for the organization of water use which can be rational only under the understanding of processes occurring in water basins. To explain the phenomenon of suppression of the transverse mixing, which is interesting and important from the point of view of ecology, a wide range of hypotheses is proposed, up to the negation of turbulence in rivers. One of the possible mechanisms for explaining the suppression of transversal mixing can be the presence of transverse circulation manifesting itself as Prandtl’s secondary flows of the second kind. The characteristic velocity of these circulation flows is very small and difficult to measure directly by instruments; however, in our opinion, they can significantly complicate the transverse mixing at the confluence. The proposed hypothesis is tested in computational experiments in the framework of the three-dimensional formulation for dimensions of a real water object at the mouth of the Vishera River where it meets the Kama. Calculations demonstrate that, at sufficiently large flow rates, the two waters practically do not mix in the horizontal direction throughout the depth over long distances from the confluence. It has been found that a two-vortex flow is formed downstream the confluence, which just attenuates the mixing; the fluid motion in the vortices is such that, near the free surface, the fluid moves from the banks to the middle of the riverbed.