This paper studies environmental corruption via a random matching evolutionary game be- tween a population of firms and a population of bureaucrats who have to decide whether to release a “green” license to the firms. A firm obtains the license if the bureaucrat checks that it complies with environmental regulations, otherwise it is sanctioned. The model assumes that there are two types of bureaucrats (honest and dishonest), two types of firms (com- pliant and non-compliant), and two possible crimes (corruption and extortion). Corruption occurs when a dishonest bureaucrat accepts a bribe from a non-compliant firm, while extor- tion occurs when a dishonest bureaucrat claims a bribe from a compliant firm. When there is no dominance of strategies, we show that there exist two bistable regimes, in which two attractive stationary states exist, and two regimes with an internal stable equilibrium, corre- sponding to the mixed strategy Nash equilibrium of the one-shot static game, surrounded by closed trajectories. From comparative statics analysis performed on the latter two dynamic regimes, it emerges that policy instruments may help the Public Administration reduce both corruption and extortion, although increasing sanctions and detection probability do not al- ways get the desired results.