Organisations managing recreational sites commonly need to understand the factors influencing visitation choices made by the public and the impact they have on the value of their sites. This need is particularly pertinent with an increasing societal reliance on cost benefit analysis for project appraisal. Whilst on-site visitor surveys can provide information on preferences and values, the potential to transfer findings to predict visitor numbers and values at unsurveyed sites provides an attractive policy option. Indeed, the demand for these benefit transfer methodologies is increasing as more Government emphasis is placed on evaluating the economic potential of rural outdoor recreation. This research concerns the development of benefit transfer models to estimate visitor numbers from outset zones to British Waterways and Forestry Commission sites. Employing a GIS, the research uses multilevel statistical modelling techniques to quantify the impacts of the proximity to competing recreation sites, resource accessibility and quality, and the characteristics of visiting populations. The models are constructed using visitor survey data and applied to unsurveyed sites, testing their use in benefit transfer. Methods are also developed that allow their output to be used to estimate the non-market value of the recreational opportunities afforded by the resources. The findings show some robust relationships determined visit patterns, with travel times from outset zones being a consistent predictor of visitor numbers. A range of other indicators were also significant including socio-demographic measures, site characteristics and substitute availability values. Nevertheless, when individual sites were compared, considerable variability was detected in the strength and direction of these relationships. The methodology developed explicitly addresses the frequently ignored spatial dimension of benefit transfer. Here the GIS provides the functionality to produce a range of measures of the underlying determinants of recreational visits. Although further refinements are needed, the future for spatial benefit transfer models appears promising.