Research on nonpoint pollution control instruments has focused primarily on incentives applied either to production inputs that affect nonpoint pollution, or to ambient pollution concentrations. Both approaches may in theory yield an efficient solution. However, input-based incentives will generally have to be second-best to make implementation practical. Design issues include which inputs to monitor and the rates to apply to them. The limited research indicates that second-best, input-based incentives can be effective in adjusting input use in environmentally desirable ways. Alternatively, ambient-based incentives have theoretical appeal because efficient policy design appears to be less complex than for input-based incentives. These incentives have no track record nor close analogues that demonstrate potential effectiveness, however. Research on how households and firms might react in response to ambient-based incentives is needed before these instruments can be seriously considered.