If agricultural subsidies are largely capitalized into farmland values then expanding support for agriculture may not benefit farmers who rent the land they farm. Suddenly reducing subsidies may be problematic to the extent that land values already embody expectations about future subsidies. Existing evidence on the incidence of subsidies on land values is mixed. Identification is obscured by unobserved or imprecisely measured factors that tend to be correlated with subsidies, especially land quality and time-varying factors like commodity prices and adverse weather events. A problem that has received less attention is the fact that subsides and land quality on rented land may differ from owned land. Since most farms possess both rented and owned acreage, farm-level measures of subsidies, land values and rental rates may bias estimated incidence. Using a new, field-level data set that, for the first time, precisely links subsidies to land parcels, we show that this bias is considerable: Where farm-level estimates suggest an incidence of 20 to 79 cents of the marginal subsidy dollar, field-level estimates from the same farms indicate that landlords capture just 10–25 cents. The size of the farm and the duration of the rental arrangement have substantial effects. Incidence falls by 5–15 cents per acre when doubling total operated acres, and the incidence falls by 0.1–1.2 cents with each additional year of the rental arrangement. Low incidence of subsides on rents combined with the farm-size and duration effects suggest that farmers renting land have monopsony power.