The emergence and prevalence of antibiotic resistance genes (ARGs) in food pose a great threat to public health and have attracted globally attention. Shotgun metagenomics sequencing is widely used to characterize ARGs in the environment. However, the distribution, co-occurrence patterns, and host information of ARGs in raw milk to the cold chain transportation duration with shotgun metagenomics sequencing, are not fully understood in China. In this study, species annotation and ARG annotation were performed on 40 raw milk samples collected from Beijing (BJ), Hebei (HB), Inner Mongolia (NMG), Shanghai (SH), and Guangdong (GD) using shotgun sequencing based on the Illumina platform. A total of 4731 bacterial species were annotated, of which 37 were dominant (relative abundance > 1%). And also, 259 ARGs were annotated, corresponding to 69 kinds of antibiotic resistance. The NMG samples, which had the lowest bacterial community richness and bacterial diversity, were annotated with the highest number of ARGs, revealing the potential risk of bacterial resistance in NMG samples. Spearman correlation analysis and network analysis were showed that the probable bacterial hosts of multiple highly abundant ARGs (MacB and VanRC) in samples from provinces except for the BJ samples were Pseudomonas and Serratia, rather than conventionally known foodborne pathogens. It is noteworthy that the relative abundance of Pseudomonas fluorescens conformed to the Allometricl function with the cold chain transportation duration, the finding that could provide a reference for assessing the freshness and the risk assessment of antibiotic resistance genes of raw milk. In addition, the dairy industry should be aware of the risk of ARGs transferring from highly abundant commensal bacteria. Raw milk for transportation by cold chain, higher abundance of Pseudomonas in some samples should be kept in mind, and the roles of ARGs carried by these bacteria should be further investigated.