In this talk we discuss molecular mechanism for transport of a small solute molecule across an interface between two immiscible liquids. The considered model system consists of hexadecane, water, and a spherical solute. The goal of this study is to develop a Langevin equation for the solute transport. In this model, the solute is modeled explicitly, whereas the effect of the solvents on the solute motion is represented by an effective mean force and a thermal bath. It is usually assumed that the fluctuations of the thermal random force are adequately described by the white noise, i.e. that the correlation time of the random force is much smaller than the characteristic time of the solute transport. We demonstrate that although this assumption is correct when the solute is located sufficiently far from the interface, the correlation time of the random force becomes significant within a very narrow (less than 1 nm wide) region of the interface. We demonstrate that the slow fluctuations of the random force in this narrow region are caused by density fluctuations of the two fluids in an area surrounding the solute. Unlike the random collisions of the solute with the solvent molecules in homogeneous fluids, the density fluctuations at the interface change the composition of the fluid surrounding the solute. This process is relatively slow and hence leads to the slow force fluctuations. We further show that the region of the slow force fluctuations lies within the area of a large gradient of the potential of mean force acting on the solute, i.e. in the area of large resistance to the solute transport. Therefore, the long-time force correlations play a significant role in the solute transport. Once the connection between the density fluctuations and the random force is established, we propose a modification of the transition-state theory for the solute transport to account for the long correlation times of the random force in the areas corresponding to large resistance to the solute transport. In conclusion, we present evidence that similar phenomena are observed in a large class of interfacial systems, including surfactant-covered oil-water interfaces and lipid bilayers.