Can roads contribute to forest transitions?

New roads are widely seen to aid economic development. New roads have also been shown to cause deforestation, suggesting development-environment tradeoffs for the locations where studies have focused to date. Yet in other settings, multiple mechanisms could support a different road-forest relationship, at least partially alleviating this tradeoff. New roads could promote reforestation by: (1) raising the relative productivity of labor in non-agricultural sectors, thereby reducing agricultural activity; (2) facilitating price convergence across regional forest-product markets, raising profits from forest management or plantations; and/or (3) encouraging substitution from locally collected fuelwood to other energy sources. India’s Rural Roads Program offers an opportunity to explore these hypotheses in a large country with little previous investigation of the road-forest relationship. Program rules prioritized construction by village population ranges, which differentiated construction timing across villages. I exploit this with a generalized difference-in-differences estimation strategy, having combined satellite, survey and census data to create a village-level, countrywide panel for 2000–2014. Road impacts are found to have considerable variation across economic settings within India: frontier-like settings saw reductions in tree cover due to new roads, while less isolated settings with greater existing agricultural development saw increases in tree cover. These results inform the spatial targeting of roads for environmentally-sensitive development, while broadening the set of mechanisms used to explain “forest transitions” (the reversal of forest cover loss).