Temporal variability is a major source of uncertainty in current life cycle assessment (LCA) practice. In this paper, the recently developed dynamic LCA approach is adapted to assess freshwater ecotoxicity impacts of metals. The objective is to provide relevant information regarding the distribution and magnitude of metal impacts over time and to show whether the dynamic approach significantly influences the conclusions of an LCA. An LCA of zinc fertilization in agriculture was therefore carried out. Dynamic LCA is based on the temporal disaggregation of the inventory, which is then assessed using time-horizon-dependent characterization factors. The USEtox multimedia fate model is used to develop time-horizon-dependent characterization factors for the freshwater ecotoxicity impact of 18 metals. Mass balance equations are solved dynamically to obtain fate factors as a function of time, providing both instantaneous (impact at time t following a pulse emission) and cumulative (total time-integrated impact following a pulse emission) characterization factors (CFs). Time-horizon-dependent CFs for freshwater ecotoxicity depend on the emission compartment and the metal itself. The two variables clearly influence metal fate aspects such as the maximum mass loading reaching freshwater and the persistence time of metals into this compartment. The time needed to reach the total impact for each metal may exceed thousands of years, so the time horizon used in the analysis constitutes a determining factor. The case study reveals that the results of a classical LCA are always higher than those obtained from a dynamic LCA, especially for short time horizons. For instance, at the end of a 100-year fertilization treatment, only 25 % of the impacts obtained through traditional LCA occurred. Results show that dynamic LCA enables assessing freshwater ecotoxicity impacts of metals over time, allowing decision makers to test the sensitivity of their results to the choice of a time horizon. For the particular case study of zinc fertilization over a period of 20 years, the use of time-horizon-dependent CFs is more important in determining the dynamics of impacts than the timing of emission.