<p>Crude oil pollution is a significant global environmental challenge. The eastern Gansu Province on the Loess Plateau, an important agricultural region containing the Changqing Oilfield, is facing increasing crude oil contamination. Understanding how microbial communities respond to varying pollution levels is critical for developing effective bioremediation strategies. This study examined how different concentrations of crude oil affect soil properties and microbial communities in Qingyang City, eastern Gansu Province, China by comparing lightly polluted (1895.84–2696.54 mg/kg total petroleum hydrocarbons (TPH)), heavily polluted (4964.25–7153.61 mg/kg TPH), and uncontaminated (CK) soils. Results revealed that petroleum contamination significantly increased total organic carbon (TOC), pH, C:N:P ratio, and the activities of dehydrogenase (DHA) and polyphenol oxidase (PPO), while reducing total nitrogen (TN), available nitrogen (AN), total phosphorus (TP), available phosphorus (AP), available potassium (AK), soil organic matter (SOM), soil water content (SWC), the activities of urease (URE) and alkaline phosphatase (APA), and microbial alpha diversity (<i>P</i>&lt;0.050). Light pollution (LP) soils demonstrated an increase in culturable microorganisms, whereas heavy pollution (HP) soils exhibited increased hydrocarbon-degrading microbes and higher expression of key functional genes, such as alkane monooxygenase (<i>AlkB</i>), cytochrome P450 alkane hydroxylases (<i>P450</i>), catechol 2,3-dioxygenase (<i>C23O</i>), and naphthalene dioxygenase (<i>Nah</i>) (<i>P</i>&lt;0.050). Non-metric multidimensional scaling (NMDS) and redundancy analysis (RDA) indicated evident variations in microbial community structure across different oil contamination levels. LP soils were dominated by bacterial genera <i>Pseudoxanthomonas</i> and <i>Solimonadaceae</i>, whereas <i>Pseudomonas, Nocardioides</i>, and hydrocarbon-degrading genera (<i>Marinobacter, Idiomarina</i>, and <i>Halomonas</i>) were predominant in HP soils. The fungal genus <i>Pseudallescheria</i> exhibited the most pronounced abundance shift between LP and HP soils (<i>P</i>&lt;0.050). Environmental factor analysis identified AN, SWC, TN, SOM, and alpha diversity indices (Shannon index and Chao1 index) as the key differentiators of CK soils, whereas the pollutant levels and metal content were characterized in HP soils. Hydrocarbon-degrading microbial abundance was a defining trait of HP soils. Metabolic pathway analysis revealed enhanced aromatic hydrocarbon degradation in HP soils, indicating microbial adaptation to severe contamination. These findings demonstrated that crude oil pollution suppressed soil nutrients while reshaping the structure and function of microbial communities. Pollution intensity directly affected microbial composition and degradation potential. This study offers valuable insights into microbial responses across contamination gradients and supports the development of targeted bioremediation strategies for oil-contaminated loess soils.</p>