Microbial drinking water quality in premise plumbing systems (PPSs) strongly affects public health. Bacterial community structure is the essential aspect of microbial water quality. Studies have elucidated the microbial community structure in cold tap water, while the microbial community structures in hot tap and shower water are poorly understood. We sampled cold tap, hot tap, and shower water from a simulated PPS monthly for 16 consecutive months and assessed the bacterial community structures in those samples via high-throughput sequencing of bacterial 16S rRNA genes. The total relative abundance of the top five most abundant phyla (Proteobacteria, Actinobacteria, Bacteroidetes, Cyanobacteria, and Firmicutes) was greater than 90% among the 24 identified phyla. The most abundant families were Burkholderiaceae, Sphingomonadaceae, unclassified Alphaproteobacteria, unclassified Corynebacteriales, and Mycobacteriaceae. A multiple linear regression suggests that the bacterial community diversity increased with water temperature and the age of the simulated PPS, decreased with total chlorine residual concentration, and had a limited seasonal variation. The bacterial community in hot tap water had significantly lower Shannon and Inverse Simpson diversity indices (p < 0.05) and thus a much lower diversity than those in cold tap and shower water. The paradoxical results (i.e., diversity increased with water temperature, but hot tap water bacterial community was less diverse) were presumably because (1) other environmental factors made hot tap water bacterial community less diverse, (2) the diversity of bacterial communities in all types of water samples increased with water temperature, and (3) the first draw samples of hot tap water could have a comparable or even lower temperature than shower water samples and the second draw samples of cold tap water. In both a three-dimensional Non-metric multidimensional scaling ordination plot and a phylogenetic dendrogram, the samples of cold tap and shower water cluster and are separate from hot tap water samples (p < 0.05). In summary, the bacterial community in hot tap water in the simulated PPS had a distinct structure from and a much lower diversity than those in cold tap and shower water.