Beach recreation value is an important consideration in a cost-benefit analysis of coastal development or conservation. A destination choice-based travel cost analysis is often used to quantify recreation values but the destination choice only partially reflects the intrinsic characteristics of that site. Visitors are influenced by opportunities available at other sites and can visit multiple sites resulting in spatially correlated errors. For this study about the recreation value of beaches on the Coromandel Peninsula we draw on the theory of cumulative attraction to analyse the compatibility of different beaches for the multiple-destination visitors who comprise more than half our sample. We investigate different random utility model formulations to explain destination choice and find that a cross-nested logit with sites nested by availability of amenities explains the observed patterns of visitation well and is more computationally efficient that non-closed-form models such as mixed logit. We also include inverse distance variables to the nearest amenity of each type and their significance supports the tenet of cumulative attraction that the importance of other spaces is greater when the attributes differ. Overall beach recreation values are maximised when sites are diverse in terms of development level and type.