In order to better understand the factors that influence bacterial diversity and community composition in moss-associated bacteria, a study of bacterial communities in four moss species collected in three seasons was carried out via high-throughput sequencing of 16S rDNA and 16S rRNA. Moss species included Cratoneuron filicinum, Pylaisiella polyantha, Campyliadelphus polygamum, and Grimmia pilifera, with samples collected in May, July, and October 2015 from rocks at Beijing Songshan National Nature Reserve. In total, the bacterial richness and diversity were high regardless of moss species, sampling season, or data source (DNA vs. RNA). Bacterial sequences were assigned to a total of 558 OTUs and 279 genera in 16 phyla. Proteobacteria and Actinobacteria were the two most abundant phyla, and Cellvibrio, Lapillicoccus, Jatrophihabitans, Friedmanniella, Oligoflexus, and Bosea the most common genera in the samples. A clustering algorithm and principal coordinate analysis revealed that C. filicinum and C. polygamum had similar bacterial communities, as did P. polyantha and G. pilifera. Metabolically active bacteria showed the same pattern in addition to seasonal variation: bacterial communities were most similar in summer and autumn, looking at each moss species separately. In contrast, DNA profiles lacked obvious seasonal dynamics. A partial least squares discriminant analysis identified three groups of samples that correlated with differences in moss species resources. Although bacterial community composition did vary with the sampling season and data source, these were not the most important factors influencing bacterial communities. Previous reports exhibited that mosses have been widely used in biomonitoring of air pollution by enriching some substances or elements in the moss-tag technique and the abundant moss associated bacteria might also be important components involved in the related biological processes. Thus, this survey not only enhanced our understanding of the factors which influence microbial communities in mosses but also would be helpful for better use and development of the moss-tag technique in the environmental biomonitoring.