Mapping the distribution of the quantity and value of forest benefits to local communities is useful for forest management, when socio-economic and conservation objectives may need to be traded off. We develop a modelling approach for the economic valuation of annual Non-Timber Forest Product (NTFP) extraction at a large spatial scale, which has 4 main strengths: (1) it is based on household production functions using data of actual household behaviour, (2) it is spatially sensitive, using a range of explanatory variables related to socio-demographic characteristics, population density, resource availability and accessibility, (3) it cap- tures the value of the actual flow rather than the potential stock, and (4) it is generic and can therefore be up-scaled across non-surveyed areas. We illustrate the empirical application of this approach in an analysis of charcoal production in the Eastern Arc Mountains of Tanzania, using a dataset comprising over 1100 obser- vations from 45 villages. The total flow of charcoal benefits is estimated at USD 14 million per year, providing an important source of income to local households, and supplying around 11% of the charcoal used in Dar es Salaam and other major cities. We discuss the potential and limitations of up-scaling micro-level analysis for NTFP valuation