We can observe new reasons for improving water protection policies: climate change calls for higher security for drinking water sources, increasing demand for improving water quality for swimming purposes in recreational areas, higher pressure on responsibility for water quality crossing national borders. Our paper presents an application of combinatorial auction theory to the problem of cleaning waters in streams where individual polluters can create various coalitions when submitting proposals for financial support to individual or common projects. The practical case has shown significant savings of social costs if the coalition solutions are realized. There is an information asymmetry between the polluters when elaborating common application for an investment subsidy for their projects. A series of economic laboratory experiments was carried out to pretest subjects’ potential behaviour here. The results showed that 75% of the groups found the (theoretical) first best solution. Subsidies required tended to be relatively low, making the competition for subsidies very stiff.