Climate modelling, to a great extent, is based on simulating air-sea interactions at larger scales. Small-scale interactions and related phenomena, such as wind-generated waves and wave-induced turbulence are sub-grid processes for such models and therefore cannot be simulated explicitly. In the meantime, the waves play the principal role in the upper-ocean mixing. This role is usually parameterised, mostly to account for the wave-breaking turbulence and to describe downward diffusion of such turbulence. The main purpose of the paper is to demonstrate that an important physical mechanism, that is the ocean mixing due to waves, is presently missing in the climate models, whereas the effect of this mixing is significant. It is argued that the mixing role of the surface waves is not limited to the mere transfer of the wind stress and energy across the ocean interface by means of breaking and surface currents. The waves facilitate two processes in the upper ocean which can deliver turbulence to the depths of the order of 100m directly, rather than diffusing it from the surface. The first process is due to capacity of the waves to generate turbulence, unrelated to the wave breaking, at all depths where the wave orbital motion is significant. In the present paper, the concept of the wave-induced turbulence is discussed, and laboratory measurements of such turbulence are presented. The second process is Langmuir circulation, triggered by the waves. Such wave-controlled mixing should cause seasonal variations of the mixed-layer depth, which regulates the thermodynamic balance between the ocean and atmosphere. These variations are parameterised and introduced in a climate model with a purpose of reproducing the pre-industrial climate. Comparisons are conducted with the NRL global atlas, and it is shown that as a result of wave mixing meteorological extremes are significantly enhanced.