Percolation is a probabilistic model of wetting of porous medium, the spread of blight in an orchard, a forest fire, etc. Specifically, bond percolation is defined by giving the occupied and vacant states with probability and , respectively, for each edge on a graph. It is known that phase transitions occur at a critical point  in this model, and it is believed that some quantities exhibit power-law (critical phenomena). For example, it is predicted that the susceptibility (the mean cluster size) asymptotically behaves like . The exponent particularly takes the value  in high dimension, which is called a mean-field value. In this talk, I will explain the basic topics of mean-field behavior for percolation models. I will also mention the infrared bound and the lace expansion. They are key topics in my research.