Groundwater contamination at active petrol stations is a global concern, yet distinguishing natural hydrogeochemical backgrounds from anthropogenic signals remains challenging. We conducted a comprehensive survey at 21 petrol stations in central China, analyzing 62 groundwater and 21 soil samples for 45 parameters, including metals, BTEX, MTBE, petroleum hydrocarbons, and conventional ions. Using Positive Matrix Factorization (PMF) coupled with spatial statistics and hydrogeological analysis, we resolved four robust sources. Factor 1 (geological background) contributed 60–70% of Mn/Al and 48–93% (mean 68%) of the individual soil heavy metals. Factor 2 (reductive pollution) contributed 75–85% of NH₄⁺-N, oxygen consumption (COD_Mn), and elevated Fe, exceeding 90% in certain hotspots. Factor 3 (petroleum hydrocarbons) contributed 75–85% to MTBE (mean 83%) and 50–65% to groundwater petroleum hydrocarbons (C₁₀–C₄₀; mean 58%), confirming point-source releases. Factor 4 (industrial solvents) dominated VOCs at an isolated site (92%). Conventional exceedance assessment misclassified geogenic Mn (26.2-fold exceedance, 62% natural) as high priority while underestimating anthropogenic NH₄⁺-N (moderate mean exceedance, 81% human-derived). The “coarse-over-fine” stratigraphy controls spatial patterns, with low-permeability clay inducing long natural-attenuation lags (8–>30 years). We propose a source-differentiated management framework, transitioning from compliance-based to risk-based regulation. This approach provides a transferable paradigm for source apportionment in urbanizing regions facing similar contamination challenges.