In 1993, Slovic developed “asymmetry principle” of trust by contrasting trust degree of college students to positive and negative events, that is to say, negative events has a much stronger effect on decreasing social trust than positive events on increasing it; “trust is much easier to destroy than to create.” He thought it was because of some negative psychological tendencies. Siegrist and Cvetkovich(2001) called them “negativity bias”. White et al.(2003) proposed the other two explanations: extremity bias and confirmatory bias. Over the past ten years, different researchers replicated Slovic’s classic study, continually and deeply explored asymmetry principle by different types of information and hazard, as well as pre-attitude. Their results partly supported asymmetry principle. They concluded that asymmetry would be influenced by many issues like pre-attitude. In certain conditions, trust would be symmetry. Therefore, they pointed that asymmetry was not general attribute of trust, but special representation on some conditions. Earle、Cvetkovich、Siegrist etc refuted asymmetry principle by framework of trust, suggesting that trust included social trust and confidence. The former was based on shared values, obeying similarity principle, thus it was resilient, symmetry; the latter was based on objective behavior standards. Once events were incongruent to these standards, confidence would be destroyed, thus it was fragile, asymmetry. Earle et al. concluded that, previous researches mainly aimed at confidence, resulting in asymmetry principle. They considered that social trust was the key point of studies on trust. This paper pointed that, the reason why Slovic developed “asymmetry principle” of trust is that, he didn’t differentiate social trust and confidence; and he chose nuclear power which is high risk hazard and public couldn’t accept at all. Thus, any information about nuclear power would affect public trust, especially negative ones. The researchers followed replicated and expanded Slovic’s classic study. Though they got some valuable results on conditional asymmetry, they were still based on asymmetry principle and negativity bias, and didn’t notice the construction of trust. Thus their contribution is limited. Earle et al. suggested that trust was symmetry because they were aware of the definition or structure of trust, aiming at social trust rather than confidence. And they analyzed it by three models——SVS Model, TCC Model, and Function Model. Their work promoted theoretical exploration of trust. In a word, the reason why there was contradiction like “asymmetry” and “symmetry” was that, there were differences on the definition or framework of trust. The former saw trust as unidimensional rather than multidimensional developed by the latter. No matter trust is symmetry or asymmetry, in real risk management, we should make great efforts to quest for measures to improve public social trust and confidence, promote the development of technology, enhance the level of life, and stabilize risk management.