Depending on the policy area and agency, bureaucratic agencies have more or less discretion to implement policy. This discretion can come in two forms. It can come in the form of insulation from congressional control. Or it can come in the form of a wider range of policies from which the agency can choose. One way that Congress affects this level of discretion is by using administrative procedures, such as whether the agency is independent of existing bureaucracies, whether the agency is headed by a board, and how many appointed officials there are. These administrative procedures could affect the amount of influence politicians can exert over an agency or they can affect the range of options available to the agency. That is, they can affect either the mean or the variance of policy implementation. This paper uses an empirical strategy that allows us to observe the effect of administrative procedures on both the mean and the variance of policy. Using distributive program awards between 2983 and 2005, we test whether agencies with more discretion target program awards more in accord with the preferences of their political overseers in Congress. We also test whether agencies with more discretion are able to choose from a wider range of policies by assessing whether the variance in distributive spending, conditional on the modeled mean, is greater among more insulated agencies. We find that administrative procedures affect discretion by reducing the variance in agency decisions, not by making them more responsive to the electoral needs of members of Congress.