How should we understand deontic statements, for example, "Ken ought to go to bed now"? Such a statement is different from a usual descriptive statement, such as "Ken is going to bed now." The ought statement is not confirmed by the observation that he is going to bed now. What is the meaning of a deontic statement, and what, if anything, makes a deontic statement correct or incorrect? Philosophers have been asking these questions. This dissertation proposes an unconventional approach to them: the best imperative approach. The best imperative approach takes the correctness or truth of a deontic statement, for example, "Ken ought to go to bed now", to be understood in terms of the bestness of a piece of advice (instruction, order, suggestion, or demand) in the imperative mood, such as "Ken, go to bed now." This relation explains why deontic statements are intuitively similar to imperatives, peculiarly action-guiding and conduct-coordinating in many contexts. In addition, given this relation, if we understand what makes certain imperatives the best, we can also know what makes the corresponding ought statements correct. What makes an imperative the best? The prescribed action must be most conducive to a certain end, and it must also be practicable. The dissertation thus seeks a theory of conduciveness and practicability, and considers its implications for the nature and evaluation of ought statements, for example, moral ought statements. The guiding methodological idea is that the theory should account for and systematize the practice of evaluating imperatives and ought statements and our intuitions about it. Among other things, it turns out that the correctness of ought statements varies with the context they are given, but there must be some reason for the addressees to comply with correct ought statements. Further, generally speaking, the action prescribed by an ought statement is not evaluated by itself; it is rather evaluated in combination with compliance with other prescriptions. In addition, there are two different ways of evaluating practicability, one of which concerns personalized action guidance, while the other of which concerns setting up shared standards for a certain type of agents.