In nonpoint source pollution problems, the regulator does not observe each polluter’s individual emission, which prevents him from using the conventional policy instruments. Therefore, new instruments have been designed to regulate this type of pollution. In an experiment, we compare the efficiency of some of these instruments: an input-based tax, an ambient tax/subsidy, an ambient tax, and a group fine. We assume that polluters themselves are affected by environmental damages. A control session without any regulation is also carried out in order to study the “status quo” situation. Our experimental data show that the input tax and the ambient tax are very efficient and reliable, and the group fine is fairly efficient and reliable. These instruments improve social welfare with respect to the status quo. On the contrary, the ambient tax/subsidy decreases social welfare with respect to the status quo, and its effect is very unreliable.