The shedding of severe acute respiratory syndrome coronavirus (SARS-CoV) into saliva droplets plays a critical role in viral transmission. The source of high viral loads in saliva, however, remains elusive. Here we investigate the early target cells of infection in the entire array of respiratory tissues in Chinese macaques after intranasal inoculations with a single-cycle pseudotyped virus and a pathogenic SARS-CoV. We found that angiotensin-converting enzyme 2-positive (ACE2 ؉ ) cells were widely distributed in the upper respiratory tract, and ACE2؉ epithelial cells lining salivary gland ducts were the early target cells productively infected. Our findings also have implications for SARS-CoV early diagnosis and prevention.Severe acute respiratory syndrome (SARS) is a recently emerged human infectious disease caused by a zoonotic coronavirus (SARS-CoV) that has a mortality rate of approximately 10% (11,20,24,30). Although it is known that SARSCoV is transmitted via saliva droplets, the source of high viral loads (up to 6.38 ϫ 10 8 copies/ml) in patients' saliva remains elusive, especially during the acute phase of viral replication (41). We hypothesized that viral replication in the upper respiratory tract may contribute to the rapid viral shedding into saliva droplets. To address this hypothesis, it is necessary to investigate SARS-CoV infection at the site of viral entry, including the identification of early target cells.Our current understandings of the target cells for SARSCoV are largely based on autopsy specimens of patients who died of SARS. Multiple cell types in the lower respiratory tract were found to be infected, including type I alveolar epithelium, macrophages, and putative CD34 ϩ Oct-4 ϩ stem/progenitor cells in human lungs (2,7,10,13,28,29). These results, while very informative, do not address the question of which cell types support the initial seeding of infection in the upper respiratory tract. Since the SARS outbreak in humans has subsided and because many of the current scientific questions are difficult to address in humans, several animal models have been developed.Animal models that have been used to study SARS-CoV infection include mice, hamsters, ferrets, cats, and nonhuman primates (cynomolgus macaques, rhesus macaques, common marmosets, and African green monkeys) (1,12,14,22,26,27,32,33,35). Although none of these animal models reproduce lethal SARS, most support SARS-CoV infection to some extent and therefore have contributed greatly to attempts to develop vaccines and therapeutics (1,9,14,19,43). We have recently demonstrated that rhesus angiotensin-converting enzyme 2 (ACE2), the primary receptor of SARS-CoV, supports viral entry as efficiently as does its human homologue (3, 23). Moreover, Chinese-origin rhesus macaques (Macaca mulatta) experimentally infected with a pathogenic SARS-CoV developed lung damage similar to that seen in humans with SARS (31). The Chinese rhesus macaque model, therefore, offers us an opportunity to study the early target cells of SARS-CoV in the resp...