Attempts to evaluate the economic benefits associated with complex environmental policies, using the contingent valuation (CV) method, have been dogged by controversy. In particular, debate has centred on the influence of embedding and related effects on the validity of CV willingness-to-pay (WTP) estimates. This paper discusses these effects in the context of identifying the most appropriate WTP elicitation treatment to evaluate the willingness to pay for the various elements of a multi-dimensional environmental policy. The findings of an empirical experiment demonstrate that a top-down allocation treatment, which uses independent sub-samples for individual policy components, provides the most reliable treatment to value all aspects of a multi-dimensional policy, whilst also overcoming embedding bias problems.