The environmental issues have drawn much attention in the studied offshore industrial park since its starting in 1999. Many high-potentially polluted plants were accommodated in the studied offshore industrial park. Soil and groundwater is the ultimate containment of environmental pollutants, and soil and groundwater contamination may directly threat to the surroundings and public health. This study utilized statistical diagnosis technology to anatomize groundwater quality data from 19 water quality measurements of 73 monitoring wells in Yulin offshore industrial park. Principal component analysis is able to derive the potential sources affecting groundwater quality, and cluster analysis is able to delineate the spatial domains of each potential sources. The integration of principal component analysis and cluster analysis is capable of achieving the forecasting of groundwater monitoring and intensifying the management mechanism of groundwater quality.By using principal component analysis, 19 water quality measurements in the original data set were extracted into 6 principal components (PCs). The first PC was recognized as salinization factor, which the characteristics of groundwater salinization is ascribed to the reclamation by using sea sand. The second PC was recognized as rust pollution factor, which resulted from the release of Zn, Ni and Cu during pipeline rust recovery processes. The third PC was identified as mineralization factor, which the most abundant Fe and Mn in the earth shell might gradually dissolve in groundwater. The fourth PC was defined as organic pollution factor, which is likely released from the leakages in the manufacturing processes. The fifth PC was defined as ammonia pollution factor, which the sources of ammonia might include agricultural operation, fishery activity, and industrial leakages. The sixth PC is identified as nitrate pollution factor, which was the nitrification product of ammonia.In order to correspond to 6 PCs, 73 monitoring wells were classified into 6 groups by using cluster analysis. Based on the locations of monitoring wells, the domains of 6 groups were spatially allocated. 66 monitoring wells located at the regions of tanks and pipelines were classified into cluster 2, which was corresponding to the potential release of Zn, Ni and Cu during pipeline rust recovery processes in the second PC. The other clusters contained only 1 monitoring well, which were corresponding to the identified source in the individual PC. Statistical diagnosis technology was proved its feasibility of identifying potential sources, allocating the affecting domain of potential sources, and improving groundwater quality management.