This paper investigates the impact of food prices on children’s cognitive development by exploiting historical price and census data in the mid to late 19th century and early 20th century United States. I explicitly model the relationships among food prices, nutrition, and cognitive development for both non-farm and farm households and use the model to motivate my empirical strategy. My empirical results confirm that there exist statistically significant differences between the two types of households in terms of the partial effects of food prices on children’s cognitive development. Using the preferred specification of this paper, I find that on average, a 1% increase in food price level reduces children’s probability of literacy by 0.44% for non-farm households and 0.37% for farm households; the average food price effect for farm households is 5/6 of that for non-farm households, after controlling for nonfood prices, household wages, demographic characteristics, household environments, and agricultural production inputs. These results send an important message to policymakers who want to address childhood nutrition and cognitive skill issues in developing countries—policy prescriptions need to take the population composition into consideration.