<p>Intensive maize cultivation in the Brantas River basin, East Java, Indonesia, has increased the use of atrazine and phosphate-based fertilizers, raising concerns regarding drinking water safety. This study evaluated the performance of a nanofiltration system for atrazine and total phosphate removal, assessed the influence of coagulant type (Al₂(SO₄)₃ and FeCl₃) and biofiltration as pretreatment units, and optimized operational conditions using Response Surface Methodology (RSM). Raw water underwent coagulation followed by slow sand biofiltration prior to NF. Membrane fouling was evaluated by monitoring permeate flux decline and further characterized through SEM–EDX analysis of the membrane surface morphology and elemental composition. Statistical evaluation using Design-Expert Version 13 (DX13) indicated that operating time, coagulant type, and biofilter inclusion significantly affected contaminant rejection (p &lt; 0.0001). The optimum configuration, determined at a desirability value of 0.779, consisted of FeCl₃ coagulation without biofiltration, operating for 9.828 days. Under these conditions, atrazine and total phosphate rejections reached 74.21% and 59.79%, respectively, closely matching model predictions. Mechanistic interpretation suggests that FeCl₃ hydrolysis species enhance atrazine removal through complexation and adsorption, while phosphate removal is governed by charge-neutralization and ligand-exchange interactions. Fouling analysis confirmed inorganic–organic deposition as a key operational constraint. Although technically feasible, scale-up requires improvements in recovery rate, membrane configuration, and fouling control to ensure long-term sustainability. This integrated assessment provides practical and mechanistic insights into optimizing nanofiltration-based treatment for micropollutant-impacted river water.</p> Graphical Abstract <p></p>