In this paper we consider the theoretical and empirical ramifications of welfare measurement in revealed preference models when weak complementarity does not hold. In the context of a Kuhn–Tucker model of recreation demand we show that, while it is possible to estimate preferences that do not appear to exhibit weak complementarity, the calculation of welfare measurements from these models requires a cardinal interpretation of preferences that cannot be tested. Furthermore, we reiterate the under-appreciated fact that even traditional use value estimates require a cardinal restriction on preferences that, while often intuitive, also cannot be tested. We demonstrate empirically that the choice of restrictions can have significant ramifications, as use value estimates can vary based on the assumed preference structure.