We collect data on prices, travel costs and farmer decisions to quantify market access for chemical fertilizer and its impact on agricultural productivity in 1,180 villages in Northern Tanzania. Villages at the bottom of the travel cost-adjusted input price distribution face 40-55% less favorable prices than those at the top. A standard deviation increase in village-level remoteness is associated with 20-25% lower input adoption. A spatial model of input adoption conservatively estimates that total trade costs are 4X pecuniary travel costs. Counterfactuals suggest that halving travel costs would more than double adoption and reduce the adoption-remoteness gradient by 63%.