自從「三鹿奶粉事件」之後，食品安全在中國大陸受到空前高度 的重視。由此，頒布了新的《食品安全法》，並在全國建立食品污染 物和食源性疾病監測条統。然而，在處理食品安全方面的個別突發事 件時，各相關政府機構的表現卻難以令公眾滿意。如瀋陽「毒豆芽」 事件反映出政府部門之間的責任推諉，表明多頭分段式的政府監管模 式將食品安全屬性予以硬性切割。本文認為權力碎片化已成為食品安 全領域公共行政需解決的首要難題。政府組織雖以專業化分工提升行 政效率，但在職權與分工方面的組織邊界隔閡、關係對立和結構衝突，使得政府管理職能陷於四分五裂的窘境。本文主要探討中國食品安全 監管權力碎片化的獨特歷史原因，相關社會背景和與權力分散、功能 割裂、部門行政改革策略之間的關聯。藉著提出由整體性治理理論作 為權力碎片化的管理工具，進而實現中國食品安全監管本土化的整體 性整合，以達到食品生產全過程的無縫接軌的監控目的。

Since the Sanlu Milk Powder Incident, food safety has attracted a great deal of attention in Mainland China. A new food safety law was promulgated. A monitoring system for food pollution and foodborne diseases has been set up all over the country. However, the ways the relevant government authorities have dealt with food safety incidents have been inadequate and have not satisfied the public. For example, the Poisonous Bean Sprout Incident in Shenyang showed that regulatory authorities were primarily concerned with putting the blame on other parties. This indicates that the multi-agency and segmentary management model of the government hinders the efficiency of responses to food safety incidents. We believe that fragmentation of power has become the primary problem in the public administration of food safety. Though bureaucratic organization may improve administrative efficiency through specialization, the boundaries between organizations, institutional rivalries, and structural conflicts have hindered the way the government functions. This essay explores the historical causes of the fragmentation of power of the Chinese government in food safety regulation, and the relationship among government units, the decentralization of power, the disruption of functions, and strategies for reforming government administration offices. We propose that the holistic governance theory should be adopted as a tool to deal with the fragmentation of power, with the aim of integrating localized agencies in food safety supervision and regulation in China. The aim is to achieve seamless regulation in the whole process of food production and consumption.