国际救灾理念发生转变，风险认知的研究日益受到重视。我国洪水灾害严重，开展洪水灾害的实证研究具有重要的现实意义。政府救灾是国家防灾减灾体系中的重要组成部分，公众对政府救灾的信任，会对其风险认知和响应行为产生影响。本文从公众层面入手，在洪水多发的长江流域选取九江、巢湖两地为具体调查样区，通过问卷调查与入户访谈相结合的方法获取数据，采用频数统计、赋分计算、对比分析、卡方检验等方法，研究了公众对政府救灾的信任水平，并分析其影响因素，以及信任对公众水灾风险认知的影响。发放问卷1020份，回收有效问卷931份，主要运用excel2003和spss15.0软件进行数据统计和分析，得出以下结论：1．政府提供的救灾措施与公众实际需要存在一定偏差，对精神层面的救助缺乏重视。公众对政府灾后救灾款物的发放表现出“不够了解、一般满意”的态度。公众对政府的信任度是社会基本防灾减灾措施中最低的（防洪工程＞自身减灾能力＞监测预警＞政府救灾）。2．公众对政府救灾的信任水平，受到个体属性因素与区域本底差异的共同影响，表现为学历越高、家庭月收入越多、受灾次数越少、灾害损失程度越轻、地区经济发展水平越高，公众对政府越信任。说明信任具有区域共构的特性，受区域经济发展和灾害风险水平的影响。3．对政府救灾的信任，不是公众判断水灾风险的主要依据，因此对其风险认知的影响不明显，公众对水灾的担心程度和发生可能性的判断都较高。同时，信任对公众备灾减灾行为倾向的影响不明显。调查区公众普遍具有较强的风险意识，但主要源于过去经验的累积而非对水灾事件的科学认识，表现为行为上明显倾向于灾害发生前后的应急减灾行为，而对长期的日常备灾行为缺乏重视。4．信任是有动机性的，信任动机的出发点不同，导致信任的结果、效应不同。科技减灾（监测预警、防洪工程）和管理减灾（政府救灾）由于作用时段和减灾原理的不同，导致公众产生了不同的信任动机，以及不同程度的科技信任和管理信任，它们对公众的风险认知的影响各不相同，后者的影响较少。因此，政府救灾工作要提高工作效率、加强监督力度、增加工作透明度，同时扮演好引导者、组织者的角色，并针对不同地区、不同群体特点的受灾公众采取不同的工作方式，以提高公众对政府工作的信任水平，这也是建设和谐稳定社会的需要。

As the concept of disaster relief changes in the world, the study of disaster risk perception has become more and more important. It is significant to study flood disaster in China because of the serious damage. The government disaster relief is the important part of national disaster prevention and mitigation system and the trust in it can influence the public’s risk perception and emergency response. The paper started from the public level, selected Jiujiang in Jiangxi Province and Chaohu in Anhui Province along the Yangtze River which was suffered from serious floods in the history as the study areas, and adopted measures of frequency statistics, scores calculation, comparison and significance test in order to study the public’s trust in local government disaster relief, impact factors and its influence on risk perception and emergency response. There are 931 pieces of effective questionnaires, of which the data were analyzed by excel 2003 and spss15.0. The result shows as follows.1. There are deviation between the disaster relief measures offered by government and the needs of the public. Spiritual level measures are lack of attention. The public doesn’t know enough and has general satisfaction with the distribution of disaster relief funds and materials. The public trust in government is the lowest in the social fundamental disaster prevention and mitigation measures(flood control engineering>disaster relief ability of their own > flood early warning>government disaster relief)2. The trust level on the government disaster relief is influenced both by the sample’s attributes and the local background attributes. The trust level is higher as the public has higher education background, more income, less disaster experience, lighter damage and more developed regional economic.3. The trust in government disaster relief doesn’t significantly influence the public’s risk perception because trust is not the main judgment. The public has highly evaluation about flood occurrence possibility, while the trust doesn’t significantly influence people’s response action tendency. The public in the study areas generally has strong sense of risk but it mainly comes from past experience accumulation instead of scientific understanding on flood disasters, so the public prefers emergency disaster relief measures to long-term disaster preparation measures.4. The outcomes of trust are different because the motivations of trust are different. Science and technology disaster relief measures(flood early warning and flood control engineering ) and management disaster relief measure(government disaster relief ) have different acting period and principle which cause different trust motivation and influence people’s risk perception.Therefore, the government should increase work efficiency, strengthen supervision，increase operational transparency, be care of public feeling and adopt different measures according to public attribution in order to increase people’s trust in government disaster relief, which is also important to social stability.