Standard models of practical rationality face a puzzle that has gone unnoticed: given a modest assumption about the nature of deliberation, we are apparently frequently briefly irrational. I explain the problem, consider what is wrong with several possible solutions, and propose an account that does not generate the objectionable result.