Deforestation studies have attracted considerable attention over the past two decades. Analyses of local deforestation examine the significance of numerous explanatory variables for the extraction of forest products. This study distinguishes among the explanatory variables as household-specific and site-specific, and argues for in-depth analysis of household-specific variables. It examines the role of household size, income, and education on firewood extraction and beedi-making in the Western Ghats of India, a hotspot of biodiversity. The study documents the incidence of these livelihoods, investigates the interaction among the three variables, and uses logistic regression models to determine the exact probability for the livelihoods. Two novel features of the study include analysis of wood and non-wood extraction through a single dataset and determination of model by the data. The results show high dependency of households on forest-based livelihoods, situations when the probability of livelihoods changes substantially, and nonparallel trends for household dependency. The study suggests the role of household-specific variables as the agents of motivation and persistence for forest-based livelihoods and presents a hypothesis about the critical size of household in forest dependency. The study recommends four measures-harvest of ecosystem services, rise in income, migration and mobility, and environmental education-to promote sustainable livelihoods. © Springer Science + Business Media, B.V. 2008.