In this study, we investigate the effect of feed concentration and recovery on the rejection of pesticides with NF/RO membranes. There is a discrepancy between laboratory studies carried out at relatively high pesticide concentrations and low recoveries and real-life conditions where filtration is carried out at lower pesticide concentrations and high recoveries, but this is often neglected when evaluating membrane rejection properties. Four commercially available polymeric thin film composite nanofiltration (NF) and reverse osmosis (RO) membranes were used to remove two phenoxy acid herbicides (2-methyl-4-chlorophenoxy acetic acid (MCPA) and 2-methyl-4-chlorophenoxy propionic acid, MCPP) and a pesticide transformation product (2,6-dichlorobenzamide, SAM) from three groundwaters of different geochemical origins in Denmark. To elucidate the prevailing rejection mechanisms, experimental data was compared to modelled rejections based on steric hindrance. Feed concentration of pesticides showed a considerable negative impact on the rejection of the charged phenoxy acids by negatively charged NF membranes, although for RO membranes the effect was negligible, showing that for NF membranes and charged pesticides, the rejection must be evaluated at actual concentrations. For all three pesticides, increasing recovery was found to lead to increased rejection values, which was ascribed to membrane pore blocking effects.