The paper proves a necessity of changing the approach to anti-corruption expertise. The analysis of opportunities for mala fide behavior of agents and evaluation of incentives for their bona fide behavior must be supplemented by the assessment of proposed regulation quality. In the paper two different algorithms of the extended anti-corruption expertise have been introduced: first algorithm is applied to the new regulation tool, second one – to the regulation tool which has been used in the past and some information on agents’ reaction is available. In the both cases the expertise starts from the modelling of society’s preferences and comparing them with the principal’s preferences which are modelled on the base of proposed regulation. The relationship between proposed algorithms of anti-corruption expertise and the typology of principal-agent models, based on the assumptions of bona /mala fides of the Principal and the Agent, is underlined. The algorithm of extended anti-corruption expertise of first type is