This paper proposes and illustrates the use of a new approach to benefit transfer for the non-market valuation of environmental resources. It treats transfer as an identification problem that requires assessing whether available benefit estimates permit the parameters of a preference function to be identified. The transfer method proposed uses these identifying restrictions to calibrate preference parameters and bases the benefit estimates on that preference function. The approach is illustrated using travel cost, hedonic and contingent valuation estimates, as well as combinations of estimates. It has three potential advantages over conventional practice: (1) it allows multiple, potentially overlapping estimates of the benefits of an improvement in environmental quality to be combined consistently; (2) it assures the transferred estimates of the benefits attributed to a proposed change can never exceed income; and (3) it provides a set of additional "outputs" that offer plausibility checks of the benefit transfers.