There is a large consensus in the literature on the negative impact of corruption on economic growth and development. Due to its illegal and covered nature, the analysis of corruption is challenging to economists. The lack of data is the main barrier for researchers. Experimental approach allows to generate the required data and could become the most promising approach to study the determinants of corruption and to test the possible anti-corruption measu­res. In this paper we present the wide survey of the experimental evidence of corruption. In the first part we make the brief introduction into the experimental economics approach. The second part reviews in details the game-theoretic сorruption models. We find out that the lion’s share of these models has a standard design based on the simple two-agents’ interaction through bribe­ry. In the third part we discuss the main findings encountered by the literature that use experimental techniques to study corruption. In particular, we analyse the cultural and gender difference in the propensities to engage into corruption behaviour, income and monitoring effects, factors that influence one's tolerance of corruption. Finally, in the last part we discuss advantages and disadvantages of the experimental approach for corruption studies.