Since reemergence of SARS from an animal virus could happen at any time, an understanding of the molecular evolution of the virus causing the past global epidemic may help to control future SARS outbreaks. Most of the human cases at the beginning of the SARS epidemic were caused by exposure to market animals (zoonotic source) [217]. Although more than 10 mammalian species, but no avian species, were discovered to be susceptible to infection with either SARS-CoV or SARS-related CoVs (SARSr-CoVs), both the number of animals traded in Guangdong, China, and the detection rate were higher in Himalayan or masked palm civets (Paguma larvata) than in other animals. Moreover, there were close matches between sequences of civet viruses and sequences of human viruses from each human outbreak, including the 2002–2003 epidemic and the 2003–2004 episode. These findings suggested that palm civets are important intermediate hosts for transmission of the virus to humans [218], probably through direct/indirect contact or inhalation of contaminated materials/droplets. Screening of SARSr-CoVs in palm civets by real-time RT-PCR and nested RT-PCR for detection of the N gene and P gene revealed the presence of the virus in rectal and/or throat swabs [100]. Major genetic variations in the S gene were found by genomic sequence analyses of civet viruses (SARSr-CoVs) and human viruses (SARS-CoVs), indicating that changes in the S gene are likely to be critical for shifting the virus from civet-to-human to human-to-human transmission that caused the 2002–2003 epidemic [214]. Kan et al. [100] analyzed all available SARS S gene sequences and found 27 signature nucleotide variations (SNVs). Based on SNVs and a phylogenetic tree of the SARS S genes from animal and human viruses, the viruses were divided into four groups. (i) Viruses without SNVs in the S gene, called a prototype group, were isolated from raccoon dogs and a palm civet but not from humans, suggesting that they can cause only animal-to-animal transmission. (ii) Viruses with two to seven SNVs generate up to six aa changes at the positions of 147, 228, 240, 479, 821 and 1080. These viruses were isolated from palm civets and from mild symptomatic patients (so-called low-pathogenic group) during the 2003–2004 episode, indicating that the virus in a palm civet can acquire the ability to infect humans. (iii) Viruses with 17 to 22 SNVs cause further eleven aa changes at the positions of 360, 462, 472, 480, 487, 609, 613, 665, 743, 765, and 1163. These viruses were isolated from palm civets and raccoon dogs in 2003 and from patients with severe symptoms (so-called high-pathogenic group) who had close contact with infected animals or patients in the early-phase epidemic (16 November 2002 to 30 January 2003). This evidence indicated the possibility that animal species other than civets may be intermediate hosts transferring the animal virus to humans. The SNVs of viruses in this group indicated that the virus in palm civets and raccoon dogs can evolve not only to infect humans but also to spread from one human to other humans by close contact, indicating the possibility that these animals and humans may share similar receptor structures. (iv) Viruses with 25 to 27 SNVs cause up to four aa further changes at the positions of 227, 244, 344, and 778. These viruses in this group were isolated from patients with severe symptoms in the middle phase (beginning on 31 January 2003: hospital phase) and late phase (beginning on 21 February 2003: hotel phase) of the 2003 epidemic that was responsible for the global outbreak, so-called large epidemic outbreak group [100].