This article purports to examine the nature of the burden of proof and its allocation rules. Based upon the premise that proof in litigation is a process of producing evidence to persuade the judge by the applicable standard of proof, this article analyzes the interrelation between the burden of proof and the standard of proof and argues that the burden of proof concept contains not only the burden of production but also the risk of non-persuasion. Contrary to the traditional continental burden of proof theory, this article argues that the burden of production is not merely ancillary to the burden of persuasion but has its own meaning and purpose. In order to accommodate the burden of production and the burden of persuasion, this article proposes a new theory of differentiating the burden of proof by litigation phases. This article criticizes the traditional theory’s unreasonableness to allocate the burden of proof without differentiating the burden of production and the burden of persuasion. The articles argue that the allocation of the burden of persuasion is a function of the substantive law while the allocation of the burden of production should take procedural factors into consideration. The theory of differentiating the burden of proof by litigation phases not only can regulate parties’ procedural conduct regarding producing evidence but also can clearly explain how the substantive policies and procedural factors work to determine the different allocations of the burden of production and the burden of persuasion. Moreover, the proposed theory provides a framework to accommodate the concepts of the duty of elucidation, spoliation of evidence and prima facie evidence, which are developed in recent years to deal with the asymmetrical distribution of evidence problem in the so-called modern litigation. Finally, the new theory can find strong support under Article 277 in the amended Code of Civil Procedure in 2000 and works to accomplish the new Code’s goals by regulating parties to conduct litigation in good faith and narrow the factual issues.