Accepted: 15 Jul 2013International audienceA unified model for the kinetics of O3 and *OH with NOM was proposed, calibrated and validated based on large experimental data sets. Single-phase batch experiments were done on 11 water samples from seven resources. Seasonal variations were studied on three resources. Effects of reaction time with ozone, ozone dose, pH, temperature, radical scavenger adding, and NOM dilution were studied. The experiments represented more than 1200 and 900 concentration measurements, respectively, for ozone and pCBA (*OH tracer). Mechanistic models were used for ozone self-decomposition and carbonate species kinetics. Results showed that the proposed model is robust and can handle different water characteristics and different experimental conditions: 75% of the experiments were modeled satisfactorily (for ozone and pCBA). Next, the domain of validity was determined: 6 ≤ pH ≤ 8; 1 meq.L−1 ≤ alkalinity ≤ 6 meq.L−1; 0-0.5 mgC.L−1 ≤ TOC ≤ 3.1 mgC.L−1. Only water samples with high organic (TOC > 2.4 mg.L−1) and low inorganic contents (alkalinity < 0.3 meq.L−1) could not be modeled adequately. Seasonal comparisons showed that the quality of the predictions decreases only for pCBA when having calibrated the model at another season. The model gave good results when using only 6 single batch experiments for calibration