This study explored characteristics of households affected by deforestation in their fuelwood and non timber forest product (NTFP) collections in Kampong Thom Province, Kingdom of Cambodia, where tropical lowland forests are decreasing in size owing mainly to agribusiness development and farming. Deforested areas were delineated by satellite image interpretation of forest cover change. A questionnaire survey was conducted for 161 households in six villages within three districts having varying degrees of forest cover changes per capita over a period of 5 years. Generalized linear mixed models were used to analyze characteristics of households affected by deforestation in their fuelwood and NTFP collections. The characteristics vary with the collection activity examined, either fuelwood or NTFP collection. We revealed that deforestation notably affects villagers whose non-forest fuelwood sources are scarce. People who collected fuelwood in forests after deforestation are more likely to be affected if the remaining forest area per capita near their village is less. In contrast, for NTFP collection, the size of the deforested area per capita near villages is more important than that of the remaining forest area, particularly to those who depend on NTFP collection as one of their main livelihood activities before deforestation. In contrast with much of the literature that has stated that the poor depend more on NTFPs, our results show that the relationship between household wealth and NTFP dependence was more than a simple negative correlation; i.e., the local people's material wealth was not a strong determinant in our NTFP model. The importance of stakeholder analysis and consultation with local residents in impact assessments and natural resource management involving land-use changes has been progressively acknowledged in the academic literature. In this context, we argue that it is essential to identify potentially affected groups among local residents in terms of their fuelwood and NTFP collections in a short-term period widely distributed at the provincial level. The combination of our findings, method applied and existing demographic survey networks in Cambodia provides an approach of identifying the affected residents in a transparent manner. (C) 2015 Elsevier Ltd. All rights reserved.