The presence of Naturally Occurring Radionuclides (NORs) in geothermal fluids is a recurring issue in geothermal projects all over the world. During the first pumping tests of the production well MOL-GT-S1-01 at the Balmatt geothermal plant (Mol, Belgium) in 2015, the geothermal water was transferred towards a nearby basin. The original geothermal fluid was characterized by a226Ra concentration of ~170 Bq L-1 (February 2019). In the meantime, the water in the basin was exposed to multiple cycles of rainfall and evaporation. In June 2019, the concentration of 226Ra in the basin water was ~24 Bq L-1. As the concentration of the water is still too high to be discarded as non-radioactive waste water, finding a solution to remove the radium from the basin water is required. In order to achieve this objective, two resins, namely the MnO2-PAN resin and the TK101 resin, were tested for their sorption capacity to remove the 226Ra by using 133Ba as proxy in batch-type experiments performed in the laboratory. The TK101 resin did not show a sufficient sorption capacity for 133Ba and is therefore not considered to be useful to remove 226Ra from the basin water. The MnO2-PAN resin however, showed promising results. A sorption yield of 86% was reached at pH 9.2 after a contact time of one week. At high pH, some co-precipitation of 133Ba with carbonate minerals comprising significant amounts of Ca2+ and Mg2+ was observed. Total sorption was higher at pH = 9.2 (maximum sorption) than at the pH of the basin water (pH = 7.5). Even at high Ca2+ concentrations in the solution (i.e. 5 g L-1), there was almost no competition of Ca2+ for Ba2+ sorption sites on the resin. The modelling of the sorption edge experiments with the MnO2-PAN resin showed that mainly Cu and Mn were competitive for the sorption sites at low pH values (2.0 ≤ pH ≤ 4.5), while at acid to neutral pH values (4.5 ≤ pH ≤ 7), Zn and Ni became more competitive. At high pH values (pH ≥ 7.0), Cd started to compete with Ba for the sorption sites of the resin. The modelling of the kinetic experiments showed that these results could best be described by a pseudo second-order kinetic model of the form dq/dt = k2(qe - qt)2.