By using Biolog technique, this paper studied the diversity of carbon source metabolism of microbial community in three farmland soils with different levels of arsenic (As) in an As mining area, and the relationships between the diversity of the carbon source metabolism and the main soil chemical properties. The total N (TN), total P (TP), total K (TK), organic matter (OM), total Cu, and total Zn contents in the three soils were in the order of medium level As > high level As > low level As. The average well color development (AWCD), Shannon index (H'), Simpson index (D), and McIntosh index (U) of soil microbes were significantly higher in the soils with medium and high levels As than in the soil with low level As. The principal component analysis and the fingerprints of the physiological carbon metabolism of the microbial community in the three soils demonstrated that the microbes in the soils with medium and high levels As had a significantly higher (P<0. 05) utilization rate in carbohydrates and amino acids than the microbes in the soil with low level As. The correlation analysis showed that the AWCD, H', D, and U were significantly positively correlated with the contents of soil TN, TP, OM, total As, total Cu, total Pb, and total Zn (P<0.05), and the canonical correspondence analysis revealed that soil TP, OM, total Pb, total Zn, TN, and pH rather than soil total As were the main soil chemical properties affecting the carbon source metabolism of soil microbial community. It was considered that soil nutrients could be the main factors affecting the community structure and functions of soil microbes in long-term contaminated soils.