Theory: Several models of distributive politics predict a role for parties in determining the allocation of federal outlays. Hypotheses: The number of Democratic voters will be positively correlated with federal outlays, even after controlling for demographic and socioeconomic variables. The degree to which a program will be skewed to Democrats will be a function of the amount of variation in program benefits across districts, whether the program is administered by formula, and the extent of one-party control when the program is initiated. Methods: Regression analysis of district-level data on election outcomes and federal assistance programs for the period 1984-90. Results: The number of Democratic voters is an important predictor of the amount of federal dollars flowing to a district. Programs with a greater amount of variation across districts are more heavily skewed to Democrats, as are programs administered by formula. Programs initiated in the latter half of the 1970s, a time of solid Democratic control, exhibit the greatest bias towards Democrats; programs started in the Reagan era show no such bias. Our results are consistent with a model in which parties in the United States play an important, but limited role in determining the distribution of federal dollars: given enough time, parties can target types of voters, but they cannot easily target specific districts.