For the purpose of heavy metals’ spatial-temporal trends and source allocation in arable soil with the rapid urbanization in Western China, samples were collected in two stages (2008 and 2017) in Chengdu city which was chosen for the case study. Positive Matrix Factorization (PMF) receptor models and Multivariate statistical analysis were used to understand the heavy metals’ spatial-temporal variability. The results showed that Cd, Cr, and As in arable soil were presented with an increasing trend during the 10-year period. Semi-variation analysis showed that the block basis ratios of the five heavy metals (Pb, As, Cr, Hg, and Cd) showed an increasing trend, which suggests that the spatial distribution of heavy metals in arable soil is more influenced by human disturbances. The source analysis shows that the enrichment of As is closely related to agricultural activities in both phases (2008 and 2017). Further source analysis showed that source of As did not change, but the contribution increased significantly. The main sources of Hg pollution changed from agricultural activities to medical equipment manufacturing, Cd changed from soil parent material sources to chemical industry waste emissions, and the sources of Pb and Cr expanded from single transportation sources to multiple sources such as road traffic and human construction. In this study, the examining of the temporal and spatial patterns of heavy metal contamination in farmland of typical rapidly developing cities in China can also provide a basis for the conservation and management of arable soil in similar areas in the context of rapid urbanization in China.