Within a contingent valuation (CV) survey ordering effects are defined where responses to a given question vary in a theoretically unanticipated manner according to the positioning of that question relative to others in the survey instrument. This paper employs a split-sample CV design to investigate the potential for such ordering effects upon stated willingness to pay for nested goods. In one ‘top-down’ design, values are first elicited for protecting the ‘whole’ of an endangered area after which values for a ‘part’ sub-area are elicited. A second ‘bottom-up’ treatment reverses the order of valuation tasks with the ‘part’ scheme valued prior to the ‘whole’. Results show that ‘whole’ and ‘part’ valuation responses are logically ordered and determined by a consistent set of factors irrespective of question order. However, variance to question order was observed in both the ‘part’ and ‘whole’ valuations. The correspondence of these results to standard and non-standard preference theories is explored and the difficulties and practical issues concerning valuing nested goods are highlighted.