Rational appraisal of agri-environmental policy requires a comparison of costs and benefits. Some fundamental features of such a comparison are explored. First, some of the indicators currently used to measure benefits or effectiveness are not proper measures of environmental effect. Indicators such as farmer attitudes and the level of participation in agri-environment schemes (AESs) are intermediate, not final, outputs. Benefits may or may not be measured in monetary terms. Where they are not so measured, the relevant methodologies are cost-effectiveness analysis (CEA) and multicriteria analysis (MCA). Where they are measured in money terms, the relevant evaluation procedure is cost-benefit analysis (CBA). However, CEA cannot answer questions about whether any agrienvironment scheme is worth pursuing: they can select only from alternative schemes given that one or more schemes must be executed. MCA can account for “do nothing” options but will diverge from CBA unless restrictive conditions are met: (a) scores on attributes must be the same, and (b) MCA weights must be the same as shadow prices. Any appraisal methodology must also account for time issues: (a) discounting, (b) selection of terminal period and post-policy scenarios, and (c) relative price effects. CBA offers a more comprehensive basis for comparison, but has uncertainties that primarily relate to the credibility of willingness to pay estimates of benefits. The use of expert rather than citizen indicator choice is discussed. The paper concludes that (a) CBA is to be preferred where credible benefit estimates can be secured, but that (b) CEA and MCA are second-best appraisal methodologies where CBA cannot, for one reason or another, be applied. Nonetheless, CEA and MCA carry with them considerable risks of inefficient decision-making...