There is a growing tendency, among central budget-makers and commentators, to argue budgets should be made for the long-term, rather than the traditional annual budget. This tendency is especially strong in the United States, where it has become virtually a conventional wisdom. This article explains, first, why that approach fits very poorly with most of the goals of budgeting. It then evaluates U.S. experience with approximations of long-term budgeting of three types: i) medium-term limits on discretionary spending, ii) the Social Security programme, and the iii) Medicare programme. That experience illustrates the reasons why long-term budgeting would not be a positive reform. They include the fantastical nature of many long-term forecasts, strong incentives for both deception and self-deception about the effects of planned budget totals, and ignoring the basic task of budgeting, which is to reconcile preferences about policy details to preferences about budget totals in a way that considers each.