A methodology for aquaculture site selection is presented using a combined hydrodynamic/nutritional model to predict fish farm wastes and dispersion, using the indirect fish waste estimation method, and the numerical model MIKE 21. This method can be used as a basis for assessment of the environmental impact of fish farms. The models were applied for a virtual inland fish farm to harvest 450 tons of sea bass in the west coast of Algeria. The results show that the fish farm effluents quantities and dispersion depends on the production cycle and size, the hydrodynamic parameters, as well as the food quantity and composition. The suspended waste, nitrogen and phosphorus are respectively the main contributors to the fish farm pollution. The current speed varies from 0.005 to 0.045 m/s, while the nitrogen and phosphorus concentrations reached 2.1 mg/L, and 0.03 mg/L respectively. This study allows the Algerian authorities to decide upon delivering fish farming permissions in the studied area, by identifying the suitability of the area for aquaculture, and predicting potential fish farm pollution. The use of numerical models in aquaculture could help reducing installation, water treatment costs and the environmental impact of fish farms by contributing to the sustainability of the activity.