Recent work has shown that because of differences in the physical environment across an area the choice of an instrument for regulation of nonpoint source agricultural pollution is an empirical question. The objective of this study is an empirical comparison of an economic incentive policy (an input tax) for reduction of agricultural fertilizer application with a command and control policy. Although a first-best policy would be administered at the farm level, transactions costs practically require a single policy for the whole area. In an application to the Columbia plateau, a 300% input tax is shown to be preferred to a mandated 25% reduction in nitrogen fertilizer on financial and policy efficiency grounds.