This paper develops a congressional bargaining model studying pork-barrel spending in the presence of spillovers between electoral districts. As the political benefit of pork-barrel spending becomes more concentrated, pork-barrel spending crowds out other spending in the budget. The relationship between spillovers and spending is made ambiguous, however, by competing substitution and income effects. I take the question to a data set of US congressional districts over a period of five decades, finding that physically larger congressional districts enjoy higher levels of federal employment and military presence, holding other factors constant. If size is a good proxy for spillovers, then the evidence is consistent with the model. This inference survives instrumental variables estimation exploiting the decennial reapportionment of congressional seats across states. The theory and evidence imply that electoral boundaries which concentrate the political benefits of federal spending lead to inefficient spending decisions.Draft, October, 2003.