Abstract In this work a general, process-based mass-balance model for water contaminants for coastal areas at the ecosystem scale (CoastMab) is presented and for the first time tested for radionuclides. The model is dynamic, based on ordinary differential equations and gives monthly predictions. Connected to the core model there is also a sub-model for contaminant concentrations in fish. CoastMab calculates sedimentation, resuspension, diffusion, mixing, burial and retention of the given contaminant. The model contains both general algorithms, which apply to all contaminants, and substance-specific parts (such as algorithms for the particulate fraction, diffusion, biouptake and biological half-life). CoastMab and the sub-model for fish are simple to apply in practice since all driving variables may be readily accessed from maps or regular monitoring programs. The separation between the surface-water layer and the deep-water layer is not done as in most traditional models from water temperature data but from sedimentological criteria. Previous versions of the models for phosphorus and suspended particulate matter (in the Baltic Sea) have been validated and shown to predict well. This work presents modifications of the model and tests using two tracers, radiocesium and radiostrontium (from the Chernobyl fallout) in the Dnieper-Bug estuary (the Black Sea). Good correlations are shown between modeled and empirical data, except for the month directly after the fallout. We have, e.g., shown that: 1. The conditions in the sea outside the bay are important for the concentrations of the substances in water, sediments and fish within the bay, 2. We have demonstrated “biological,” “chemical” and “water” dilution, 3. That the water chemical conditions in the bay influence biouptake and concentrations in fish of the radionuclides and 4. That the feeding behaviour of the coastal fish is very important for the biouptake of the radionuclides.