We designed specific primers and fluorescence-labeled probes to develop real-time and conventional RT-PCR assays for detection of human coronavirus NL63 or HKU1. Subsequently, experiments were undertaken to assess diagnostic criteria such as specificity, sensitivity and reproducibility. The detection limit of the real-time RT-PCR assays was 10 RNA copies per reaction mixture. No cross-reactivity was observed between RNA samples derived from designed HCoV and other HCoV or human metapneumovirus. A total of 158 nasopharyngeal swab specimens collected from adult patients with acute respiratory tract infection in Beijing were screened for the presence of human coronavirus NL63 and HKU1 by using real-time RT-PCR and conventional RT-PCR method. The fluorescence quantitative RT-PCR method detected six specimens positive for human coronavirus NL63, five specimens positive for human coronavirus HKU1; and conventional RT-PCR method detected three HCoV-NL63 positive and three HCoV-HKU1 positive, respectively. The convention RT-PCR products of positive samples were obtained and sequence analysis confirmed the reliability of the above methods. In summary, the real-time RT-PCR assay for HCoV- NL63 or HKU1 was more sensitive than conventional RT-PCR and with less time (less than 4 hours) for completion. It may be suitable for molecular epidemiological surveillance and clinical diagnosis for human coronavirus NL63 and HKU1.