Background: An outbreak of acute respiratory illness (ARI) occurred at a training base in Beijing. A total of 12 students were diagnosed with symptoms of ARI from August 26 to August 30, 2015. The cause of the ARI outbreak was investigated. Methods: In partnership with the local center for disease control, we conducted an epidemiological investigation on the outbreak, collecting a total of twelve pharyngeal swab specimens as well as patient demographic for the affected patients. We used multiplex real-time PCR to screen for sixteen common respiratory viruses in these samples. To isolate the causative virus, we inoculated Hep-2 cells with the human adenovirus (HAdV)-positive samples and then carried out sequencing and phylogenetic analysis of the hexon, fiber and penton genes of the isolated adenoviruses. In addition, we analyzed the whole genome of one strain isolated from the index case to identify single-nucleotide substitutions. Results: We identified ten HAdV-positive students by multiplex real-time PCR. None of the students were co-infected with other viruses. We successfully isolated seven strains from the pharyngeal swab specimens. Genetic analysis showed that the coding sequences of the hexon, fiber, and penton genes obtained from those seven HAdV strains were identical, suggesting that they represented seven isolates of a single virus strain. One HAdV isolate obtained from the index case, BJDX-01-2015, was selected for whole genome analysis. From this isolate, we obtained a 34,774-nucleotide sequence. Phylogenetic analysis showed that the genome of BJDX-01-2015 clustered with HAdV-B55 and had 99.97% identity with human adenovirus 55 isolate HAdV-B/CHN/BJ01/2011/55 (GenBank accession no. JX491639). Conclusions: We identified HAdV-B55 as the culprit of the ARI outbreak. This was the first reported outbreak in Beijing due to HAdV-B55 virus. Continuous surveillance of respiratory adenoviruses is urgently needed for understanding the epidemiological and evolutionary features of HAdV-B55 and could also find value in an epidemiological modeling approach.