Novel graphene oxide (GO)-based adsorbent embedded with epichlorohydrin (ECH) as a coupling agent and dimethylamine (DMA) as a ligand (GO-ECH-DMA) were prepared and employed for endotoxin removal from aqueous solutions. The physicochemical properties of nanocomposite were fully characterized. The model attributed to batch adsorption process was optimized employing response surface methodology (RSM) via various parameters such as pH, GO-ECH-DMA dosage, and contact time and endotoxin concentration. The p-value with low probability (<0.00001), determination coefficient (R2=0.99
) and the non-significant lack of fit (p > 0.05) showed a quadratic model with a good fit with experimental terms. The synergistic effects of the linear term of contact time and GO-ECH-DMA dosage on endotoxin removal were significant. The optimum condition for endotoxin removal was obtained at pH of 5.52, GO-ECH-DMA dosage of 21 mgL−1, contact time of 56 min and endotoxin concentration of 51.3 endotoxin units per milliliter (EUmL−1). The equilibrium was the better explained by Langmuir isotherm with the maximum monolayer adsorption capacity of 121.47 EUmg-1, while the kinetics of the endotoxin adsorption process was followed by the pseudo-second-order model. The adsorbent could be recycled with NaOH. The possible mechanisms of endotoxin adsorption were proposed by hydrogen-bonding, π-π stacking, and electrostatic interaction.