This talk is an attempt to review the main concepts of mesoscopic physics – physics of objects, which dimensions are intermediate between true macro and true micro scales. This year is declared Einstein Year and Word Year of Physics because of the 100-th anniversary of the Einstein’s Miraculous Year. One of the revolutionary papers that Einstein wrote in 2005 is devoted to the theory of the Brownian motion – motion of particles that are small enough to feel the molecular motion, and large enough to be observed individually. Einstein proposed statistical analysis of the motion of each Brownian particle and derived Diffusion equation to perform this analysis. This idea of statistical approach is so crucial for modern mesoscopic physics, that it would not be an exaggeration to start the history of the field from the celebrated Einstein’s paper. Mesoscopic (sample-specific) effects are usually rather weak in classical systems. Quantum interference enhances them tremendously. We will briefly discuss the effects of weak localization and sample-to-sample fluctuations of the conductance of disordered metals. One of the main outputs from studying these effects is the ability to determine rates of numerous inelastic relaxation times – energy relaxation, spin relaxation and quantum phase relaxation. This information is vital for the implementation of the Quantum Computer paradigm. We will discuss the current stratus of theory and experiment on the inelastic relaxation in disordered conductors.