Soil water repellency (SWR) poses a significant challenge in agriculture by reducing the soil's ability to absorb water, thereby limiting water infiltration and negatively impacting crop yields. Numerous studies have shown multivariate relationships between the expression of SWR and soil particle size and organic carbon contents. However, the biological effects on soils are often considered in isolation. In this study, we explored the relationships between SWR and various physicochemical and biological properties of 30 soil samples from ten nonirrigated, dryland farms in South Australia, using correlation analysis and Random Forest modelling. Ethanol droplet tests were used to quantify SWR, and key soil properties including clay content, total organic carbon (TOC), pH, total nitrogen, microbial activity (measured via dehydrogenase activity, DHA), and dissolved iron (dFe) were assessed. Soils were grouped into low (Group 1, mean SWR 0.5 M) and high (Group 2, mean SWR 3.0 M) SWR categories based on clear differences in measured SWR values. Significant differences between groups were observed for clay, TOC, dFe, and DHA. Random Forest analysis identified clay and pH as the strongest predictors in Group 1. In Group 2, the model didn't explain SWR variation. DHA had minimal influence on model performance.